Silver/Glass Mirrors for Solar Thermal Systems Applications

Silver/Glass Mirrorsfor Solar Thermal Systems

L! Notice

This publication was prepared under a contract to the United States Government. Neither the United States nor the United States Department of Energy, nor any of their employees, nor any of their contractors, subcontrac- tors, or their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsi- bility for the accuracy, completeness or usefulness of any’,information, product or process disclosed, or represents that its use would not infringe privately owned rights.

Printed in the United States of America

Available in print from: Superintendent of Documents U.S. Government Printing Office Washington, DC 20402

Available in microfiche from: National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 Stock Number: SERUSP-271-2293 Mirrors for Solar Thermal -terns

DE85000537 SERI/SP-271-2293 UC Categories: 62, 62a, 62b, 62c June 1985

Preface The research and development (R&D)described in this document was conducted within the US. Department of Energy’s (DOE) Solar Thermal Technology Program. The goal of the Solar Thermal Technology Program is to advance the engineering and scientific understanding of solar thermal technology and to establish the technology base from which private industry can develop solar thermal power production options for introduction into the competitive energy market. In solar thermal technology, tracking mirrors or lenses concentrate sunlight onto a receiver. The heat absorbed by the receiver is converted into electricity or used as process heat. The two primary solar thermal technologies, central receivers and distributed receivers, use various point and line-focus optics to concentrate sunlight. Central receiver systems use fields of heliostats (two-axis tracking mirrors) to focus the sun’s rays onto a single tower-mounted receiver. Parabolic dishes up to 17 meters in diameter track the sun in two axes and use mirrors or Fresnel lenses to focus radiant energy onto a receiver. Troughs and bowls are line-focus tracking reflectors that concentrate sunlight onto receiver tubes along their focal lines. Concentrating collector modules can be used alone or in a multi-module system. The concentrated radiant energy absorbed by the solar thermal receiver is transported to the conversion process by a circulating working fluid. Receiver temperatures range from loo°C in low-temperature troughs to over 1500’C in dish and central receiver systems. The Solar Thermal Technology Program is directing efforts to advance and improve each system concept through research and development of solar thermal materials, components,

A Product of the Solar Technical tnformation Program Produced by the Technical Information Branch A Division of Midwest Research Institute Solar Energy Research Institute Operated for the 1617 Cole Boulevard, Golden, Colorado 80401 U.S. Department of Energy

i and subsystems, and the testing and performance evaluation of subsystems and systems. Work is done under the technical direction of DOE and its network of national laboratories who work with universities and private industry. Together they are pursuing a comprehensive, goal-directed program to improve Performance and provide technically proven options for eventual incorporation into the nation’s energy supply. To contribute to the national energy supply, solar thermal energy must eventually be economically competitive with other energy sources. Components and system-level performance targets have been developed as quantitative program goals. The performance targets are used in planning research and development activities, measuring progress, assessing alternative technology options, and optimizing components. These targets will be pursued vigorously to ensure a successful program. This document reviews the state of the art of the design, manufacture, testing, and performance of silver/glass mirrors for solar thermal systems applications.

Acknowledgments This document was prepared by A1 Czanderna, Keith Masterson, and Terence M. Thomas of the Materials Research Branch of the Solar Energy Research Institute.

Contents

Page Introduction ...... iii Chapter 1 Overview of Silver/Glass Mirrors ...... 1 Chapter 2 Properties of SiIver/Glass Mirrors ...... 5 Chapter 3 Physics and Chemistry of SiIver/Glass Mirrors ...... 19 Chapter 4 Mirror Product Testing Procedures ...... 45 Chapter 5 Solar Mirror Manufacturing Methods ...... 55

ii Introduction

8ackg rou nd over the economic life span of the facility in which they are used, during which time the mirror may be A significant aspect of research and development degraded by storms and stressed by temperature (R&D)of solar thermal energy systems concerns the cycling, water, humidity, vapor, ultraviolet (UV) design, manufacture, testing and performance of radiation, dust, and physical abuse. mirrors for solar energy applications. Mirrors have Most mirrors have been manufactured by bonding a an essential role in various solar thermal technology thin layer of silver to glass - the glass being the applications. For these applications, mirrored sur- protective barrier, light transmitter, and, in part, a faces are used to redirect and/or to concentrate the structural support. A copper film to protect the rays of the sun. The most extensive application of silver’s back surface and layers of paint to protect the mirrors is expected to be in solar thermal systems. copper also are necessary. Most industrial and utility applications of solar thermal energy systems are to produce process heat The advantages of using glass as a superstrate are its fluids or develop other forms of energy, such as elec- clarity (high solar transmittance), low cost, smooth tricity for mechanical power, requiring concentrat- surface, physical strength, abrasion resistance, im- ing mirrors. Mirrors are needed to concentrate the pertneabilit y, resistance to soiling, ease of cleaning, naturally available solar flux in order to attain and inertness. Compared with other mirrors, those higher temperatures. made of glass and silver are preferred for their high reflectance, good specularity, durability, and resist- Mirrors are used mostly in such solar thermal sys- ance to distortion from loads. However, glass is heavy temsisubsystems as parabolic troughs, parabolic and brittle, requiring massive structural support. dishes, spherical bowls, and heliostats. Each has particular design requirements. However, the prin- Mirrors of silvered glass are fairly tolerant of mirror cipal attributes sought in the design specifications of compositional variations and/or impurities. They the “ideal” solar thermal reflector should include the can be expected to provide high reflectance as long as following: the silver remains intact and in contact with the glass. High optical performance: Retaining this integrity depends on chemical and - reflectanceitransinit tance, physical processes that take place in and between the layers of materials composing the mirrors. - specularity, - geometrical configuration, Although glass has many advantages as a super- Low maintenance (dust free), strate for silver mirrors, certain factors can mitigate its use. Glass is a relatively heavy (high density) and Low initial cost, and brittle material; therefore, it requires either stronger Long life. and stiffer mechanical support structures or thinner, more fragile layers than lighter materials. Most mirrors presently in use employ either silver (domestic and decorative applications) or aluminum Mirror manufacturing techniques can play an impor- films (automotive applications] for their reflecting tant role in the performance of glass as a mirror surfaces. Mirrors made of silver must be protected superstrate. Poorly made glass may contain bubbles, from chemical and physical deterioration of the which can cause dispersion of the light beam, dimin- silver; whereas,aluminum mirrors are more resistant ishing mirror specularity. The glass manufacturing to degradation, technique also influences mirror performance in that Since both silver and aluminum reflecting surfaces it controls surface smoothness and thickness uni- in solar applications must retain their highest reflec- formity. Float glasses have problems for solar appli- tivity for many years, the reflecting surfaces are pro- cations; their facings tend to be wavy and lack paral- tected with transparent covering materials referred lelism. On the other hand, the float glass process to as superstrates. Also, this adds to the physical allows the production of very thin glass desired for integrity of the reflector since the superstrate sup- heliostats. Polished sheet glass has the best unifor- ports the kilver. mity but is thicker than float glass. Thick glass tends to be less wavy than thin glass.

The material or materials attached to the silver (1) The age of the glass prior to its being silvered can should not detract from the silver’s reflective qual- affect the longevity of a mirror’s high optical perfor- ity, (2) should not contribute to degradation of the mance. Apparently, the surface of many glasses silver, and (3) should bond well. Moreover, all of “foam up” during storage which leads to a destabili- these properties should remain substantially intact zation of the subsequent silvering process.

iii The attributes sought in the design specifications of the “ideal” silver/glass mirror for solar thermal systems cannot be achieved simultaneously. As is the case in all areas of materials technology, the end product is a compromise between performance, durability, and cost: hopefully an optimal one.

Objective and Scope

The principal objective of this document is to review the state-of-the-art of the design, manufacture, testing, and performance of silver/glass mirrors for solar thermal systems applications. This document treats each of these topics. Chapter 1 briefly discusses the fundamentals of silveriglass mirrors. Chapter 2 presents the desired and actual mechanical and optical properties of silver as a solar reflector, of glasses as a superstrate for silver, and of silver/glass mirrors, Chapter 3 reviews the research literature on the physics and chemistry of silver/ glass mirrors, and includes a discussion of the silver deposition processes and of the interfacial and surface reactions of the layered mirror materials. Chap- ter 4 describes present testing procedures and results for solar mirrors and indicates improvements needed in future mirror manufacturing. Chapter 5 discusses a summary of the float glass manufacturing process for fabricating commercial silveriglass mirrors, and a preview of a prospective integrated process for manufacturing silver/glass solar collectors in a single manufacturing plant.

iv Chapter 1 Overview of Silver/Glass Mirrors

The use of mirrors in the solar energy industry is Receiver or extensive and varied. The most extensive application :eiver of silver/glass mirrors is for solar thermal electric or industrial process heat (IPH) systems. Most industrial and utility solar-energy-generated applications that produce process heat fluids or develop other forms of energy, such as electricity for mechanical power, require concentrating mirrors.

Mirrors for Solar Thermal Collectors

I Solar thermal mirrors are used in parabolic troughs, Figure 1-2 Parabolic Dish parabolic dishes, spherical bowls, and heliostats. The parabolic trough (Figure 1-1)concentrates the sun’s rays along a line. The parabolic dish (Figure 1-2) concentrates the sun’s rays toward a point. The Although different collectors each have particular spherical bowl (Figure 1-3) concentrates light along a design requirements, optimally, none should have rod. The compound parabolic concentrator (Figure surface irregularities that tend to scatter light and 1-4) funnels light from a large aperture into a smaller reduce system efficiency. The manufacture of curved receiver. The heliostat field employs a vast number reflectors for concentrating flux is critical since of reflectors (flat or minimally parabolic) to focus on departure from design contour can diffuse a mirror’s a central receiver, as shown in Figure 1-5. focus.

Concentrator Receiver reflective surface

mechanism Tracking ’ I 11

Figure 1-1 Parabolic Trough

Overview of Silver/Glass Mirrors 1 Tracking /, />

\- Receiver

I Figure 1-3 Spherical Bowl

Flux concentration ratios (collector area to area of will cause the rays of a beam to deviate from an ideal focal point) are ultimately limited by optical factors. path in all mirrors. Also, mirrors - even those of For line focusing the sun's rays, the maximum theo- solid silver with ultraclean surfaces - have a di- retical concentration ratio is 200; for point focusing electric layer that causes refraction. it is 40,000. Detracting from any theoretical limit are the imperfections contained in real mirrors; i.e., wavi- ness, scattering, and absorption sites, among others. These and other divergencies presently limit line concentration ratios to 60 and point-focus concentrations to about 600. Values of 100 and 5000, respectively, are thought to be achievable with improved technologies. Eventually, practical achievable temperatures are expected to reach about 800°C and 27oo0C using line- and point-focus collectors, respectively. However, present technology is limited to achieving temperatures of 30OoC and about 1000°C. Heliostats are expected to provide flux concentration ratios of up to I 1000 and, ultimately, tempefatures of about 2500'C. Figure 1-4 Compound Parabolic Concentrator A temperature as high as 1370'C is possible with present advanced technology. Receiver \ \ /

Preference for Silver Mirrors for Solar The r ma I Ap pI ica t ion s

Ideally, the perfect mirror is one for which each ray of incoming solar radiation is reflected - without loss of intensity and without deviation from the ideal state; however, no mirror is perfect. In all mirrors Heliostats \ some quantity of solar radiation from part of the Tower solar spectrum will be absorbed by the materials that I make up the reflector. Normal surface irregularities Figure 1-5 Central Reciever

2 Silver/Glass Mirrors for Solar Thermal Systems Most mirrors presently use either silver (domestic ity (generally 5 2 mrad [0.10 deg.]), durability, and and decorative applications) or aluminum films (auto- resistance to distortion from loads. motive applications) for their reflective surfaces. Mirrors made of silver must be protected from chem- Mirrors of silvered glass are fairly tolerant of com- ical and physical deterioration of the silver: whereas, positional variations and impurities. They can be aluminum mirrors are more resistant to degradation. expected to provide high reflectance without unreas- In many solar applications, however, silver is pre- onable dispersion, providing the silver remains intact ferred because of its high reflectivity; 97%-98% com- and in contact with the glass. Retaining this integrity pared to 88°/~-920/0for aluminum. Nevertheless, both depends on chemical and physical processes that materials are being tested and used in various solar take place in and between the layers of materials applications. composing the mirrors. Since both silver and aluminum surfaces in solar The basic ingredient of most glass is silica. However, applications must retain their highest reflectivity for many different glass formulas exist to provide such many years, the reflective surfaces are protected special features as durability, flexibility, special with transparent covering materials referred to as ranges of refraction and thermal expansion, dielec- superstrates. The mirrors are then called “second tric property, and transmission. Thus, some glasses surface.” Substrates are the base (bottom) materials are better than others for particular uses. Common that protect the back surfaces of the silver or alumi- window glass, for example, contains “fillers” to num. A mirror is “first surface” when the reflecting decrease its cost. One filler is iron oxide, a substance layer precedes the support. that strongly absorbs the long visible wavelengths of light. This absorption reduces the reflectivity of Although, ideally, a transparent covering should common silver-backed glass mirrors to values of typ- behave optically as if it were not present, in practice ically 82%. it reduces optical performance to some degree. Since with silver mirrors the superstrate is chosen for pro- There are other ways in which chemical composition tection of the silver from oxidation and other chemi- can affect performance. For example, although glass cal effects, the superstrate should minimize the is relatively impermeable, a certain flow or diffusion deterioration of silver’s “original” high state of re- of chemical constituents within the glass can take flectivity. Also, since the superstrate supports the place. Sometimes these constituents (such as sodium silver, that support should add to the physical integ- in common glass) can react with the deposited silver, rity of the reflector. Collectors must survive some degrading the silver’s reflectivity. Composition can difficult environmental stresses, including thermal also affect the mechanical characteristics of the glass cycling, hail, ice and snow loads, wind, and dust which can influence mirror integrity. A mirror with a abrasion. glass thermal expansion much different than silver will tend to delaminate during thermal cycling. Reflector integrity for solar applications is para- (Glass can be formulated to cover a range of expan- mount. The material or materials attached to the sion coefficients, including matchups with specific silver should not detract from the silver’s reflective metals.) In addition, most glass compositions are quality, should not contribute to degradation of the brittle and vulnerable to fracture from stress and silver, and should bond well. Moreover, all of these impact. properties should remain substantially intact for about 30 years. The manufacturing technique can play an important role in the performance of glass as a mirror superstrate. Poorly made glass can contain bubbles which Glass Mirrors can cause dispersion af the light beam, diminishing mirror specularity. In the past, most mirrors were manufactured as a construction of silver bonded to glass - the glass being the protective barrier, light transmitter, and, in part, structural support. These so-called second- surface mirrors (because the silver is bonded to the side of the glass furthest from the light source) also include a copper film to protect the silver’s back surface and Iayers of paint to protect the copper. The advantages of using glass as a superstrate are its clarity (high solar transmittance), low cost, smooth surface, physical strength, abrasion resistance, im- permeability, resistance to soiling, ease of cleaning, and inertness. Thus, glass approaches the ideal material for protecting silver. Compared with other mirrors, those made of glass and silver are preferred for their high reflectance (82%-97%), good specular-

Overview of Silver/Glass Mirrors 3 4 Silver/Glass Mirrors for Solar Thermal Systems Chapter 2 Properties of Silver/Glass Mirrors

The Ideal Solar Reflector The principal attributes sought in the ideal solar I ref 1e c t o r a re Absorbed Radiation High optical performance: Incident - reflectanceitransinit t ance, - specularity, - geometrical configuration, Low maintenance (dust free), Reflected *------Low initial cost, and Radiation Long life. Specu lar High optical performance means solar reflectance R, Reflected or solar transmittance T, values approaching unity, Radiation and a Gaussian beam spread due to nonspecularity and slope errors of less than 2 mrad in the most demanding heliostat applications. Figures 2-1 and 2-2illustrate the basic optical properties of transmit- Ideal Material: R, = 1 .O (Flat, Specular) ters and reflectors and compare ideal materials with Real Materials: R, = 0.80 - 0.96 real materials (Lind 1981).These desired properties apply to glass as a superstrate with a silver reflective coating. I Figure 2-2 Optical Properties of Reflectors (Lind 1981)

Approaches to the Ideal Reflector for Solar Incident Absorbed Appl icat ion s Radiation The preferred specifications for ideal solar reflectors Scattered can be defined easily but are difficult to achieve. The Transmitted reflector should be perfectly specular and should Radiation have a solar reflectance of unity. A substrate should be capable of supporting the reflective surface and maintaining a given geometrical configuration under all external environmental stresses. The reflector should not degrade in performance over its estimated 20-40 year lifetime. Finally, the cost per unit area of the entire reflector unit should be economically Specular accept able. U Transmitted The properties that are important for consideration Radiation in an ideal solar reflector are best subdivided into Ideal Materia!: T, = 1.0 {Flat, Specular) three classifications: [I)mechanical, (2) optical, and Real Materials: T, = 0.80 - 0.92 (3)chemical. The relevant specific properties within these three classifications are listed in Table 2-1. Definitions and examples of the mechanical and optical properties are provided in the following sec- I tions of this chapter; the chemical properties of Figure 2-1 Optical Properties of Transmitters (Lind 1981) silver/glass mirrors are discussed in Chapter 3.

Properties of Silver/Glass Mirrors 5 Mechanical Properties of Second-Surface Glass Mirrors

2-1 Glass Thickness, in. Table provides a listing of the mechanical proper- 01 0.2 0.3 04 ties pertinent to mirror materials and reflectors. Since most solar mirrors are often composite layers ~~COFtNl~GType 03j7 of several materials, the mechanical properties of --. (Fusion) both the individual materials and the composite pro- L duct are important. Generally speaking, the mechan- ‘J%-%.-ASG LUSTRAGlass ical properties of the individual materials must be SCHoTT (0.05 w/%Fe) compatible with each other for the composite product to achieve the desired set of mechanical Typical properties. Mirror Glass (Soda Lime) These mechanical properties take on different degrees of importance, depending upon the type of solar mirror; e.g., second-surface glass mirrors or aluminum mirrors. The following discussion of properties is taken from Bouquet (1979, 1980). Glass thickness and composition. Many types of glass, including soda-lime and low-iron glass, are currently being used for solar applications. Reflec- k I’ 1.47 rnm (0.058 in.) I III 1 I tance by second-surface glass mirrors varies remark- 2.5 5.0 7.5 1c ably with thickness and composition (Bouquet 1980) Glass Thickness, mm which produces a corresponding dependence of solar Adapted From Taketani (1977) reflectance as shown in Figure 2-3. Typical thicknesses of solar glass are identified by vertical lines, Note that most of the slanted lines shown at zero thickness of glass point to the high (90%)range on the for Figure 2-3 Solar Reflectance versus Glass Thickness for Typical reflectance scale+The criterion thickness selection Types of Second-Surface Glass Mirrors (Bouquet 1980) is that the glass be as thin as possible to reduce double absorption during the transmission of the incident solar radiation over as wide a spectrum as possible and still withstand the strength specifications or Properties affecting glass performance. In early stud- other requirements for a particular application. ies the fundamental physical properties of most commercially available glasses were evaluated for The chemical composition of several representative applications to solar mirrors. These properties are glasses is given in Table 2-2, and a summary of phys- summarized in Figures 2-4 through 2-8. Two specific ical properties is given in Table 2-3. It can be seen qualities required of flat, one-dimensional, and two- that the physical properties are highly composition- dimensional glass shapes in order to achieve high dependent, These properties will be presented and performance are low coefficient ofthermal expansion, discussed fully in the next section. and high thermal conductivity.

Table 2-1 Property Classifications for Solar Receivers

Chemical and Optical Mechanical Properties Environmental

Specu lari ty Flatness Delamination Oxidation Surface figure error Density Static fatigue Corrosion Hemispherical reflectance Coefficient of expansion Stress corrosion Solarization Solar-weighted hemispherical Thermal conductivity Crack growth Weather ing reflectance (p2) Abrasion resistance S pr i ng- bac k Ag g lo meration Superstrate transmittance Hardness Du ra b i Iity Outgassing Index of refraction Softening point Thermal cycling Annealing point Crazing Strain (yield) point Young’s modulus Poisson’s ratio Shear strength Bending strength

6 Silver/Glass Mirrors for Solar Thermal Systems used, including the following two approaches: (1) Hertz fractures produced by pressing a steel ball against the glass, and (2) the bending strength of circular glass plates. I I I lw 1 I I' pi: The latter test is found to be more reliable, consider- 3 LL* ing the fact that the side of float glass next to the tin bbbbb 5q.g F bath is observed to be the weakest. Since the side '6--, g3cm~~b co e CnEG Strain Point

7' 40r 0. qmpered(Jested in air) 30- x .- .I-----_'. :: 20- 6. v), 2. 3j 10- .-F 81 Y a 6- Annealed (Tested in air) a- ;,[ I I I I I I I 1 I I 10-3 10-2 10-1 1 10 102 103 104 105 106 107 *t Duration of Stress, seconds 0' I 1 I 1 I I 400 600 800 1000 1200 1400 1600 Temperature, O C Figure 2-5 Stress Time Characteristics of Glass (Bouquet 1979) 'Corning Glass Works

Hard High Alumina Glass (1720) (STONG) Figure 2-4 Viscosity - Temperature Curves (CIT 1978) Window Glass (STONG) I Two newer glasses, in particular, have physical properties that are well-suited to solar mirror appli- Soda Lime Glass (STONG) cations. These two glasses are Corning Glass Works 031 7 alurninosilicate glass and 7809 borosilicate glass. Their properties are tabulated in Table 2-4. Special Soda Potash Both glasses have high solar reflectance with deposited silver; and both have low coefficients of expan- t + Heavy Lead Glass (8870) (STONG) sion and, hence, low stress corrosion. 5000b I 1 I I I I I 100 200 300 400 500 600 700 800 Measurements of glass strength. Measurements of Degrees, C the strength of flat glass have been treated by Lewis (1976). Several measurement methods have been Figure 2-6 Young's Modulus of Various Glasses (Bouquet 1979)

Table 2-2 Analysis and Properties of Representative Glasses (Bouquet 1979)

Analysis, Percent by Weight Type of Glass Softening SO2 Modifiers Al2O3 B203 PbO TempnOC

Fused silica 99.9 - - - - 1667 96% silica (Vycor) 96.0 - - 4.0 - 1500 Borosilicate (Pyrex) 80.5 4.2 2.2 12.9 - 820

Al u m i nosi I icate 57.7 9.5 25.3 7.4 - 91 5 Soda-li me si I ica 73.6 25.4 1.o - - 696 Lead-al kali 54.0 11.0 - - 35 630

Properties of Silver/Glass Mirrors 7 opposite the tin bath is silvered by the mirror indus- I try, it follows that the glass side of the mirror is I 0.30f Soda-Lime Glass weakest. 0 28- One of the greatest limitations in using glass for parabolic dishes is its static fatigue, which can be m described by the following equation: BOROSIL 7440

t,. = I<@[-" where

I<,n = constants \-LeadLead Glass (22.3% PbO) t, = time to failure = the stress level at which failure occurs. 0 200 400 600 800 1000 ol Degrees, C In other studies, Turnmala (1976) has shown the stress corrosion coefficient 11 to be a function of the Figure 2-7 Mean Specific Heat of Glasses (Bouquet 1979) bulk glass composition. It changes little with the surface of the glass. Experiments show that the I stress corrosion resistance varies inversely with the (hernial expansion coefficient over the 2Fi°C to 30OoC range. Slow crack growth in glass that can lead to failure is frequently initiated hy edge cracks. Recent information indicates that R "factory cut" edge has better resistance to tldgt. cracking than edge processing (beveling, polishing, or other treatment). During glass shaping, glass has a tendency to spring b,ic:k to its original shape upon cooling. An initial t effort is need~dto ascertain the amount of spring- Ii11111111 -200 -100 0 100 200 3( back allowa1,le when selecting a new type of glass. A Degrees, C tentative criterion is less than a 1%dimensional I spring-back along the smallest radii of curvature. Figure 2-8 Thermal Conductivity of Glasses (Bouquet 1979)

Table 2-3 Physical Properties of Glass

Y 0 u ng's Thermal Specific Modulus Expansion* Refractive Poisson's Type of Glass Gravity g/cm3 1O3 kg/mm2 cm/cmo C Index Ratio (Ib/ff3) (lo6psi) (in./in." F)

Soda-lime 2.47 7.1 93.6 x 10-7 1.512 0.22-0,24 (154) (10.2) (52 x 10-7)

Aluminosilicate 2.52-2.64 8.8-8.9 42.1-46.1 x 1.53-1.547 0.24-0.25 (154.6-157.2) (12.5-12.7) (23.4-25.6 x

Borosi 1 icate 2.13-2.48 5.0-6.9 32-51.5 x 10-7 1.473 0.2-0.23 (132.8-1 54.6) (7.1-9.8) (17.8-28.6 x 10-7)

96% fused silica 2.18 6.9 7.6-8 x 10-7 1.458 0.19 (135.9) (9.8) (4.2-4.4 x 10-7)

Fused silica 2.2 7.4 5.6 x 10-7 1.459 0.16 (137.2) (10.5) (3.1 x 10-7)

*Over the range 0°C to 3OOOC or 32" F to 572" F Source: Corning Glass Works, Corning, New York

8 Silver/Glass Mirrors for Solar Thermal Systems Table 2-4 Typical Solar Mirrors: Preliminary Properties of Corning 0317 and 7809 Glass [Thickness - 1.5 mm (0.060 in.)]*

Type of Corning Glass Property 0317 7809 (A1u m i n osi Ii ca te) (Borosilicate)

Density, g/cm3 2.45 2.44 Index of refraction 1.512 1.509

Solar transmittance, O/O 90.9 (est.)** 91.7 t 2

Solar reflectance, O/O (a) Vacuum-deposited silver 95 * 1 95 (est.)** (b) Chemically deposited silver 94 k 1 94 (est.)** Weather (glass only), 20 years Excel lent Exce I Ien t Softening point, "C 870 750 Annealing point, "C NA 569 Strain point, "C NA 529 Strengthening capacity Yes No Expansion coefficient (a) 0" C-300" C, 10-7/0 C 88 77 (b) -30" C to +50° C, I O-7/oC 81 72 Poisson's Ratio 0.22 0.20 Young's Modulus, 102kg/mm2 68.94 76.52 (lo6psi) (10) (11.1) 'Corning Glass Works, Corning, New York **Bouquet 1980

Optical Properties of Glass Mirrors error (slope] variations typically result from local deviations in the surface normal about the ideal Definitions. The principal optical properties applica- surface shape. Thus, the surface figure error deter- ble to solar reflectors are specular reflectance (specu- mines the direction of the reflected radiation, and the larity), surface figure error, directional hemispheri- surface specularity determines the angular spread and cal reflectance, solar-weighted hemispherical re- intensity of this radiation; see Figure 2-9 for a repre- flectance, and superstrate transmittance. sentation of these two surface irregularity effects.

Two measures of surface irregularities that cause Directional hemispherical reflectance measures all loss of redirected sunlight from a solar mirror are of the reflected radiation from a surface independ- surface specularity and surface figure error. Specu- ently of its angular distribution. This measurement larity measures the scattering from surface varia- is typically recorded as a function of wavelength tions with a characteristic spacing of less than from 350 nm to 500 nm, using an integrating sphere approximately 1 mm, whereas scattering from sur- reflectorneter. The solar-averaged hemispherical re- face contour (slope] variations on a scale greater flectance value for a mirror is determined by averag- than approximately 10 mm is referred to as the sur- ing the spectral hemispherical reflectance data over face figure error. the solar spectrum. This value represents the maximum available reflected solar energy for a particular Surface specularity values (ASTM 1983)*are gener- mirror; however, depending on the angular distribu- ally associated with the fine structure of the reflecting tion of this radiation, only a portion of this reflected material, e.g., deposited silver; whereas surface figure radiation may be utilized in a solar collector.

'There is neither an accepted (ASTM, etc.) measure of specularity Laboratory measurements by Pettit and Roth (1980) nor any proposed measure with a scale from 0 to 1. When one is of the specular reflectance at 18 mrad [R(18 mrad)] established, it will probably be in units of milliradians. and the directional hemispherical reflectance

Properties of Silver/Glass Mirrors 9 large angle scattering at the mirror surface. The reflectance profiles illustrate the importance of measuring both the specular reflectance and hemi- Depicts Surface Figure Error spherical reflectance properties at the same wavelengths for all mirror materials. \ NRS NAS Superstrate transmittance is the ratio of the solar energy transmitted through a plane surface of super- --.Depicts strate in a defined wavelength band to the solar Surface energy incident on the superstrate in that band. Spew larit y Reflectance properties of various reflector materials. Of all possible metals that could be used for solar -- reflectors, only silver and aluminum have solar- "131- Actual Surface weighted spectral reflectance values above 0.90. All others, including gold, nickel, chromium, stainless a = Figure (Slope) Error steel, rhodium, and copper have reflectance values p = Specularity Measure below 0.82. The spectral reflectance properties of silver and aluminum, together with gold, measured for the metalivacuum interface are shown in Figure 2-1 1. The solar-averaged reflectance values are cal- Figure 2-9 Schematic Representation of a Mirror Surface Showing culated to be: silver, 0.98; aluminum, 0.92; and gold, the Difference Between Figure (Slope) Errors and the 0.85. These values represent the practical upper limit Mirror Specularity (Pettit and Roth 1980) of solar reflectance for these materials. The reflectance of aluminum is reduced over the solar region [RA(2v)],both measured at 500 nm, and of the solar- primarily because of an interband absorption cen- averaged hemispherical reflectance [R,[27~)]are tered at approximately 800 nm. It has been suggested illustrated in Figure 2-10 for several solar mirror that this absorption band may be eliminated by using materials. Silvered float glass (2.7 mrn thick) exhib- amorphous aluminum; this would increase the solar its the highest hemispherical reflectance [Rh(2n)= reflectance several percent (Trotter 1978).However, 0.921 at 500 nm of the five mirror materials, but has efforts to reduce this absorption have not been suc- the lowest solar-averaged hemispherical reflectance cessful and do not appear to be feasible now, [Rh(2rr)= 0.831. In most solar applications, the metal-reflecting layer The difference beween the specular and hemispheri- is protected by a transparent coating such as an cal reflectance values for these materials is due to oxide (e.g.,A120:,), glass, or plastic film. The index of

Material Rh(27~) R(18 mrad) R,(2n) 1 Silvered Glass 0 92 0 92 0.83 2 ALZAK - Parallel 0 89 078 085 3 ALZAK - Perpendicular 0 89 072 085 4 SHELDAHL Aluminized Teflon 0 87 0.82 0 87 5 3M SCOTCHCAL 5400 0 86 085 085

1 IAu Vacuum Dielectric J Ag 0.98 0.97 Al 0.92 0.88 Au 0.85 0.82

I L 1 I 500 1000 1500 2000 2: """" Wavelength, nm 10 12 14 16 18 Angular Aperture, mrad

Figure 2-10 Specular Reflectance Properties far Several Solar Mir- Figure 2-1 1 Reflectance Properties ol Silver (Ag), Aluminum (Al), ror Materials. The table lists the hemispherical reflec- and Gold (Au) as a Functlon of Wavelength for the tance [RA (2n)l and the specular reflectance at 18 Metal/Vacuum Inlerlace. The table lists the solar- mrad (R(18 mrad)] measured at 500 nm. Also listed is weighted rcfloctaiw?lor n metal/vacuum interface and the solar-weighted hemispherical reflectance [RS(21r)] a rnetal/didoc:lric: irltc!cI;icc with an assumed index of {Pettit and Roth 1980) refraction of 1 I) (I)iiffit and Roth 1980)

10 Silver/Glass Mirrors for Solar Thermal Systems refraction of most coating materials is typically 1.5 the outer surface is not parallel to the reflecting sur- through the solar spectral region. With a non- face, then multiple reflected images are formed. absorbing dielectric layer applied over the metals, the solar-averaged reflectance is reduced by 0.01 The reflecting surface itself can scatter radiation if it reflectance units for silver, 0.03 for gold, and 0.04 for has a rough surface (i.e.,figure error). In addition, the aluminum. Thus, in most applications, silver mirrors reflected beam intensity may depend upon the purity will have, at most, a solar reflectance of 0.97, while of the metal layer as well as the deposition process the solar reflectance of aluminum will be closer to (vacuum deposition, chemical reaction, ion plating]. Both the surface texture of the backing layer, which 0.88. is typically a thin metal sheet, and the lamination or In order to understand the specularity and the bonding technique that is used to attach the reflect- reflected beam intensity of various mirror materials, ing surface to the backing layer can be important in it is helpful to consider the construction of a typical affecting the specularity of the reflector. The same is reflector. The typical construction of both a front- true for the bonding of the backing layer to the sup- surface (substrate)and second-surface (superstrate) port structure. reflector are shown in Figure 2-12. In a front-surface mirror, the reflecting layer is applied to the substrate Optical properties of second-surface silver mirrors. and then overcoated with a protective film. Glass as the superstrate. (Taketani 1980) has performed solar reflectance measurements on several In a second-surface mirror, the reflecting layer is second-surface silvered mirrors produced by differ- applied to a transparent superstrate (e.g., glass or ent commercial silvering processes on the same glass plastic film) which is then bonded to a support struc- superstrate. The results, shown in Table 2-5, together ture after applying a protective layer. Also shown in with earlier measurements on bare silver (Taketani Figure 2-12 is a back-protection layer. The entire 1978),indicate that the reflectance of coatings depos- composite, shown in Figure 2-12, is termed a “reflec- ited by different commercial chemical- or vapor- tor,” while the outer layers, which include the silver deposition methods are comparable. The measure- or aluminum film, are termed the “mirror.” ments, except for the bare silver measurement, were made on a special low-iron-content float glass (3.2 Both the specularity and the reflected beam intensity mm thick) manufactured by PPG Industries. can be affected by all of the components shown in Figure 2-12. Depending upon its optical properties, Taketani (1980) also reported recent reflectance and the outer protective layer can modify the reflected transmittance measurements that were made on var- beam’s intensity at the metal/dielectric interface as ious commercial grades of glass of different thick- shown previously. In addition, the outer layer can nesses and iron content with and without silver absorb radiation and thereby reduce the solar reflec- backing. The results are shown in Table 2-6. The tance, The specularity can be affected by scattering value of low-iron-content glass for second-surface within the layer or at the inner or outer surfaces. If solar mirrors is readily apparent.

A. Superstrate Reflector 6.Substrate Reflector

Flgure 2-12 Examples of Typical Reflective Surfaces (Bouquet 1980)

Properties of SilverlGlass Mirrors 11 The solar t ransmittanr:e of glass is strongly influ- I,a Xio ra I o ri t?s (Goodyea I’ 1980). Figure 2- 1 6 shows a enc:t?cl hy the iron con t en1 and the t hickntiss, ;IS illus- t ypit:iil laser scan, Curve A, to detertiiine the surface trated in Figurt: 2-13 (Bouquet1979). The decrease in figure error. The short period is r;allc?d the “wave” transmittancc! with iron content is due to long wave- and the long periocl is called the “how.” Curve R 1t:ngt h al)sorption hy the t’errous state of iron, causing shows the wave only and was the hasis ofacccptance H greenish color in the glass. Ferrous iron exhibits a of the glass lor the heliostat. strong absorption liiincl ;it about 1000 nm (see Figure

2-14) (Coyle 1981).Removal of the iron or conversion Optical properties of first-surface silver mirrors. First - of the ferrous iron to ferric iron in the fusion glass- surface, or front-surface, silver mirrors face the making process has ht!c?n shown to improve the solar inherent problem that exposed silver is react ivc wit h trirnsiiiit t ante significantly. Figure 2-15, for exam- environmental gases (see Chapter 3). First-surface plt:, reveals ii 4.7% inc:rc?asein spectral transmittance aluminum mirrors, on the other hand, have ti natural for the Ford Mof 01’Company’s low-iron, 3-mrn float protective coaling of A1,0 However, various stud- glass conipare(I to its own normal piwriuct ion glass itts are underway to develop protective overcoatings (Gootiyear 1980).The calct~late~~rt?fler:tanct? of a sil- for silver. verrtd mirror using this glass is 89.6”1;1. St utlies of the reflectance of front-surface silvered ‘The effect 01’ iron cont c:nt l~t?canic?of consiclerahle float glass, protectively coated with siliconc resins, importance in t he rnanuI’a(;ture 01’ the httliostats for have been carried out by Dennis ;ind McC:ue (1980)at the Solar One 10-Megaw;it t Power Project at Bar- the Dow Corning Corporation. Samples were pre- stow, California. ‘I’ht! Glass Division of t he Ford pared with and without silane coupling agents that Mo t o r Co inp any in a n u lac:t ti red ii 1ow - i ro n- (:o n t en I tend to prwinote the adhesion of the resin to the me(al float glass lor those heliostats. surfac:e. Table 2-7provides the ineasured values of specularit y and solar hemispherical reflect ances; the In addition to t hr? :lass composition, Ford Motor rcf‘lectance values ranged from 94.8% to 97.8”h. How- Company also had to control the flatness of t he glass ever, in environmental exposure tests, the silvered so that proper direct ional refl(?ctancc could be I’loa t glass samples usually failed tiecause of loss of achieved with the heliostat mirrors. Flatness inea- adhesion hetween the glass substrate and the vapor- surements were made by Bat tellit Pacific Nort hwest d e p o s i t ed s i 1 v e r .

_-~

Table 2-5 Reflectance Measurements of Silvered Mirrors (Taketani 1980)

Silver Surface Being Reflectance Mirror Manufacturer Deposition Measured (Oj4

Theoretical Value (Kingslake 1965) vapor front 99 Mirrorlab (Taketani 1978) Texas chemical front Donelley, Michigan vapor second Buchmin, Ind., California chemical second 92 chemical second 92 Binswanger Mirror Co., Arizona chemical second 92 c hem i cal second 92 Carolina Mirror Corp., N. Carolina chemical second 91 chemical second 91 Gardner Mirror Corp., N. Carolina chemical second 90 chern ical second 91 chemical second 91 chemical * second 91 chemical* second 91 chem ical* second 91 Mechanical Mirror Works, New York chemical second 90 chemical second 90

“Enhanced Solaflect formula

12 Silver/Glass Mirrors for Solar Thermal Systems Table 2-6 Glass Reflectance and Transmission Measurements (Taketani 1980)

Glass Glass Manufacturer Thickness Reflectance Transmission

mm (in.) (O/O) 1

PPG Industries, Inc. 2.38 (3/32) 89 87 Clear float 3.18 (1/8) 86 86 6.35 (1/4) 79 82 Ford Motor Co. 2.38 (3/32) 91 88 3.18 (1/8) 89 87 6.35 (1/4) 81 83 PPG Industries, Inc. 3.18 (1/8) 92 88 Low-iron float LOF, Inc. Solar 90 sheet 3.18 (1/8) 94 90

Optical and Physical Properties of Code requirements such as very low optical absorption, 7809 Solar Fusion Glass outstanding stability in the weight, and elasticity when forming parabolic mirrors. Code 7809 solar fusion glass is ii product of a collabo- Glass forming and composition. The fusion process, ralive clevclopinent program ween Corning Glass bet patented by CGW, is one of several methods used to Works (CC:W) and [he Solar Energy Research Institute produce sheet glass. The advantages of this process (SERI)(Coyle and Livingston 1981; Coyle ct al. 1980). are (1) compositional flexibility, (2) capability to SERI sought this new glass to satisfy unique solar produce thin cross sections; i.e., 0.5 mm, (3) ease of changing the thickness, (4)excellent surface quality, and (5)operation in oxidizing conditions. Using current equipment, widths to 1.4 in and lengths to slightly over 3 m can be achieved (modifications to g41present equipment could allow widths up to 1.8 m]. 9 4- I The fusion piTocess is based on an “overflow pipe.” Molten glass t‘lows from the fore-hearth into one end 90- of a horizontally oriented hollow trough. An end view of this trough would show a V-shaped configu- 88- ration. The glass fills the internal cavity and over- 6a ai 0 86- +5 .- E 84- C E I- a2-

80 -

78 -

0 0.02 0.04 0.06 0.08 0.10 0.12 Wavelength, nrn Iron Level, O/O

Figure 2-14 Absorption Coefficients of Fe’2and Fe’3 in Soda-Lime Figure 2-13 Solar Transmittance as a Function of Iron Content and Glass (The shaded curve is the relative spectral solar Glass Thickness (Bouquet 1979) irradiance) (Coyle 1981 )

Properties of Silver/Glass Mirrors 13 flows both sides of the trough. Edge-located pulling rolls are used with the surface tension of the molten glass to draw the glass uniformly down the exterior 100- A\ of the "pipe" to the apex of the "V." There the glass "fuses" into a monolithic sheet. Heating and cooling *------_----* c~-cc------equipment permits the judicious control of tempera- 3mm Glass tures needed to carry the glass through the annealing and strain points. At the appropriate point in the cooling zone, the sheet is cut into required lengths and then is automatically transported to the edge- stripping and final-sizing station. Infrared Corning Glass Works developed the new glass composition (CGW-7809) specifically for solar energy ''A~'""" applications (see Table 2-8). Some of the properties of the glass are shown in Table 2-9. The glass was designed to be compatible with the fusion sheet-forming process and to be melted at high rates for low production costs. Compositions originally considered Figure 2-15 Transmittance for Low-Iron, 3-mm Glass (A) and ranged from "improved" soda-lime (float) glass to a Normal 3-mrn Glass (B) (Goodyear 1980) hard borosilicate pharmaceutical composition (CGW-7809). The low solar absorption by the new glass was obtained by reducing the total iron content, and adjusting the batch and melting conditions to oxidize all of the ferrous iron (Fe''] to ferric iron (Fe':'). F'J, 3L2 .- The levels of alumina and soda were adjusted in the composition to achieve a thermal expansion of 77 x 10-~/*cto minimize thermal expansion differences with potential support materials. The resulting CGW-7809 composition represents a suitable corn- promise among solar transmittance, thermal expansion, lifetime, and cost. Solar transmittance and reflectance. The spectral hemispherical transmittance of the CGW-7809 glass was evaluated by using an integrating sphere spec- 0 5 10 15 20 trophotometer [Coyle and Livingston 1981).The high I Distance, inches spectral transmittance of the 7809 glass differs markedly from the transmittance of the more corn- mon soda-lime float glass; see Table 2-10. The broad Figure 2-16 Flatness of 3-mm Glass as Measured by Laser Scan at 1100 BatteHe Pacific Northwest Laboratories. (Glass for absorption band around nm in the soda-lime Curve A is not corrected for bow; that for Curve B is glass due to Fe+2is conspicuously absent in the 7809 corrected for bow.) (Goodyear 1980) glass, as shown in Figure 2-17. The transmittance of

Table 2-7 Reflectances of Silvered Float Glass (Dennis and McGee 1980)

Specularity in mrad at 568 nm Solar Directional for Cone Angles Resin Hemispherical Reflectances Rh (2779 10" 45"

0.955 3.17 3.84 0.957 0.89 1.60 0.959 0.73 0.77 0.978 0.58 0.60 0.949 3.20 3.60 0.948 8.20 8.65

14 Silver/Glass Mirrors for Solar Thermal Systems coner Glass Company on 7809 glass were about 2% Table 2-8 Composition of CGW-7809 Glass lower. The 7809 fusion glass has higher transmit-. (Coyle et al. 1980) tance and provides higher reflectance than the other two glass types over the full range of thicknesses Oxide YO considered. An evaluation of 7809 glass for solarization (changes 66.0 of transmittance due to solar irradiation, particu- 9.0 larly in the ultraviolet) showed no detectable change 8.0 in optical density (less than 0.001) after 500 hours at 9.0 10 suns in a solar simulator [Coyle et al. 1980).This 2.0 result allayed concerns that solarization might reduce 5.0 the solar transmittance of the glass in its service 0.5 environment. Comparable experiments on several 0.2 float glasses (Vitko 1980) revealed a decrease in opti- 0.0 cal density (i.e., an increase in transmittance) vary- ing from 0.002 to 0.01.

Optical accuracy. The surface flatness of a glass Table 2-9 Properties of CGW-7809 Glass sheet determines the optical accuracy of mirrors in (Coyle et at. 1980) redirecting solar energy, as well as the aesthetic appeal of glazings used in passive applications. Fig- ure 2-19 shows the results of a laser-ray tracing of the 7809 glass (Coyle et al. 1980).It shows that glass Softening point, “C 750 from the pilot test reflected light less accurately than Annealing point, “C 569 float glass, probably, CGW believes, as a result of Strain point, “C 529 poor surface tension leveling of the glass on the Expansion, (O C)-l,0-300” C 77 x 10-7 Harrodsburg, Kentucky, “fusion”pipes. However, the Poisson’s ratio 0.20 0317 glass made at the Blacksburg, Virginia, full- Young’s modulus, psi 7.6 x lo3 kg/mm2 scale “fusion” facility showed optical accuracy ap- Density, g/cm3 2.44 proaching that of the best float glass - more than Index of refraction 1.509 adequate for solar applications. Thus, CGW is opti- mistic that the optical accuracy for 7809 obtained at Blacksburg would be adequate since it is expected to be similar to that of the 0317 glass. the 7809 glass is close to the theoretical maximum; it Chemical durability. Accelerated weathering studies shows little absorption, and only reflective losses. were made on 7809 glass specimens exposed at 7OoC and 100% relative humidity (Coyle et al. 1980).CGW Figure 2-18 iliustrates the solar-spectrum-weighted observed an increase of about 0.002 in. in the optical transmittance of unsilvered glass and the hemispher- density after four weeks of exposure+However, the ical reflectance of silvered glass for conventional change was an order of magnitude less than that float glass, low-iron float glass, and 7809 glass for a noticed in float glasses under similar conditions, range of glass thicknesses. It gives theoretical values indicating that the glass was durable in the environ- for reflectance. Values for mirrors silvered by Fal- ment expected for solar applications.

Table 2-1 0 Solar Transmittance of Various High-Transmittance Glasses (Coyle and Livingston 1981)

Thickness Glass Solar Glass Type (in.) Composition Transmittance 0.040 7809 fusion Alumino-borosilicate 0.91 9 0.060 7809 fusion Al um i no-borosi Iicate 0.91 7 0.060 0317 fusion Al u m i n 0si I i cate 0,909 0.090 0317 fusion Al um i nosi Iicate 0.91 0 0.110 0317 fusion Al umi nosi Iicate 0.91 0 0.1 25 Low-iron rolled Soda-lime silicate 0.91 0 0.1 18-0.125 Low-iron rolled Soda-lime silicate 0.88-0.89 0.1 25 Regular float Soda-li me si Iicate 0,83-0.87

Properties of Silver/Glass Mirrors 15 Mechanical properties. The fracture toughness KIc: and stress corrosion properties [I

7809 Fusion Glass t - 32-10'2rnr (7 g4\g3 (0 125 - 0 040 in ) Float Glass (Low Iron) t: 32-229rnm [O 125 - 0 090 lfl

Total Hemispherical Reflectance of Silvered Glass O/o

Reflectance IS a Functlon of Silvering Quality Variations Up to a 6% Have Been Noted

Figure 2-1 8 Solar Reflectance of Second-Surface Silvered Mirrors on Different Glasses (Coyle and Livingston 1981j *In units of megapascals x (meters)"'

cz 0 +so > a0 'Q c a 50 2 0 2 5,-

30(d F a Lt 0 c 0 'G 0 0 2 U

0 0 4.0 6, Angular Deviation (rrir,icl)

Figure 2-19 Fraction of Glass Surface Area with More than Angular Deviation lrorn Spwiiliir Hi*tlt*c tlon (( I 1~11~,iii(t Livingston 1981)

16 SilverlGlass Mirrors for Solar Thermal Systems Bibliography tors.” Solar Reflective Materials; Proceedings of the Second Solar Reflective Materials Workshop; Coyle, R. T., McFadden, J. D. 0. [March 1982).Stress Sun Francisco, CA: 22-14 February 3980. Appears Corrosion of a High-Transmittance Glass for Solar in Solar Energy Materials, Vol. 3 (Nos. 1,2); pp. Applications. SERI/TP-255-1507. Golden, CO: 285-300. Solar Energy Research Institute; 22 pp. (Presented Goodyear, J. I<.; Lindberg, V. L. (September 1980). at the Electrochemical Society Annual Meeting; “Low Absorption Float Glass for Back Surface Denver, CO; October 1981.) Solar Reflectors.’’Solar Reflective Materials; Pro- ceedings of the Second Solar Reflective Materials Cuddihy, E. F. (September 1980). “Theoretical Con- Workshop; San Francisco, CA; 12-14 February siderations of Soil Retention.” Solar Reflective 2980. Appears in Solar Energy Materials, Vol. 3 Materials; Proceedings of the Second Solar Reflec- (NOS.1, pp. 57-67. tive Materials Workshop; San Francisco, CA; 12- 2); I4 February 1980. Appears in Solar Energy Mate- I

Properties of Silver/Glass Mirrors 17 18 SilverlGlass Mirrors for Solar Thermal Systems Chapter 3 Physics and Chemistry of Silver/Glass Mirrors

Glossary of Terms in Chapter 3 Introduction

AES Auger Electron Spectroscopy The chemical and physical processes involved in the preparation and degradation of silver/glass mirrors e.m,f+ electromotive force are discussed in this chapter. As illustrated in Figure EPR Electron Paramagnetic Resonance 2-12 (superstrate) in Chapter 2, the silver/glass mirror consists of a multilayer stack of glass, silver, and ESCA Electron Spectroscopy for Chemical a protective backing. The protective backing may Analysis (see XPSj also consist of two or more multilayers. The most important stability issues for a multilayer configura- IR Infrared tion are interface reactions, adhesion, and the choice ISS Ion Scattering Spectrometry of the protective backing. In the descriptions of possible causes of mirror degradation in this chapter, NRA Nuclear Reaction Analysis reference is made to various analytical techniques to RBS Rutherfard Backscattering Spectrometry discern the causes. A brief description of the purpose of the techniques is contained in Table 3-1. ref. radio frequency Since most mirrors are prepared by an electroless, RH Relative Humidity wet, chemical process method [i.e., chemical-process SAM Scanning Auger Microscopy mirrors), the emphasis of this section is on the chemistry related to preparing these mirrors and the SEM Scanning Electron Microscopy processes responsible for their degradation. Similar SEXAFS Surface Extended X-ray Fine Structure but less complete discussions are provided for mirrors made by vacuum, organometallic, and other pro- Spectroscopy cesses. Recent research results on improving the SIMS Secondary Ion Mass Spectrometry adhesion in the multilayer stacks and on identifying a more durable protective backing for the silver are uv Ultraviolet then presented. In this section a brief historical XPS X-ray Photoelectron Spectroscopy (see development of silveriglass mirrors and a general ESCA) overview of parameters that degrade the performance of mirrors in multilayer stacks is presented. XRF X-ray Fluorescence

Table 3-1 Surface Analysis Techniques for Evaluating Mirror Degradation

Technique Rationale Auger Electron Spectroscopy (AES) Identify surface composition inctuding impurities and degradation products; supurb lateral resolution Scanning Electron Microscopy (SEM) Map topography and surface defects Ion Scattering Spectroscopy (ISS) Determine surface cornposition; sensitive to isotopes and the first monolayer Secondary Ion Mass Spectroscopy Determine surface composition for first few atomic layers, especially ppm detectability of impurities X-ray Photoelectron Spectroscopy Identify oxidation state, average chemical composition, and nature of chemical bond for the first few monolayers of a surface

Physics and Chemistry of Silver/Glass Mirrors 19 Historical Development and without a metallization backing, (2)at the silver/ glass interface during UV exposure, (3) of any metal- The historical development of mirrors has been lization backing with atmospheric gases and then summarized by Schweig (1973),who describes mir- with silver, (4) of impurities incorporated into the rors in use from more than 4000 years ago up to the multilayer stack during preparation, and (5) of the technology of 1973. The method for making a glass surface itself with atmospheric gases or at- chemical-process mirror has not changed in principle t a (: h ed part icu f a t es . since Lieliig’s observation (1835) that a solution of silver is reduced by aldehydes to deposit ii brilliant General Comments About Mirror mirror. When the discovery of tin chloride as a sensi- Degradation tizer was niade in 1876, the silverrd mirror was able to compete with the tin-mercury amalgam mirror for Some field-exposed, second-surface, silver/glass cla ri t y and d u 1-ati i 1 it y. solar mirrors fabricated by conventional wet chem- The general processes for making mirrors ~‘~OJTIthat istry processes have experienced significant loss in point on were almost exc:I~isiveJyused for. 80 to 100 performance because of degradation in less than one years (Schweig 1973).During this time the processes year (Mastersnn 1983a). Others have been exposed were made more economical and efficient. The optical for more than five years without undergoing visible quality improved, but the durability of the product changes (Bomar 1981). The variability in the field did not change, The description of the basic manu- durability of mirrors made by the same process is not facturing process has not changed since 1970 when fully understood. Based on prior work, implementing the process was economically optimized by intro- realistic quality assurance tests for durability and ducing formaldehyde as the spray-process plating projecting mirror lifetime from accelerated tests is accelerator. not possible. There is some evidence that the variability in durability is partly a quality control prob- Schweig (1973) also traces the worldwide develop- lem, but there is much more evidence that the basic ment efforts that have resulted in the current, widely problem is the thermodynamic instability of the used manufacturing method for preparing chemical- glass/silver/copper/paint system when exposed to process mirrors; this industrial method is described terrestrial environmental conditions (Thomas 198-). in Chapter 5. He also summarizes the detailed chemical formulations used for making chemical-process Research over the past few years identified three mirrors, as well as equipment used, safety precau- failure modes that are common to many of the field- tions, etc. The 200-page book devotes about 15 pages degraded mirrors: (1)debonding of the silver/glass to mirrors made by methods different from the chem- interface, (2)halide formation in the metallic layers, ical process. and (3)agglomeration of the silver layer (Masterson 3983a). All of these are consequences of basic rnech- anisms at work that involve halides, water, oxygen, General Parameters Causing Degradation and other impurities (Thomas 1983). Halides have been reported by numerous investigators [Czanderna The life-cycle cost of a particular mirror design 1978; Daniel 1980; Pederson 1980; Mills 1980; Shelby depends on the durability of the reflector assembly, 1980a; and Thomas 1983). Pinholes result primarily such as shown in Figure 2-12 in Chapter 2,The pur- from agglomeration; black spots result from dissolu- pose of sandwiching the reflective material into a tion of copper, leaving voids at the silveriglass inter- multilayer stack is to limit reactive constituents face (Delameter 1981). from reaching the silver; thus, increasing the performance lifetime of the mirror, Mirror durability, Early attempts at duplicating these failure modes in therefore, depends on the materials used to form the the laboratory by subjecting mirror specimens to mutilayers and the environmental exposure parame- environments containing high levels of degradative ters. The degradation problems associated with multi- conditions in order to accelerate the process have met layer stacks clearly depend on the ability of the glass with mixed success [Dake 3980; Coyle 1981a; Pelli- and protective backing to exclude reactive materials, cori 1980). The environrncnIs range from boiling the ability of the manufacturer to exclude reactive water and salt spray to acid vapor and other corro- impurities during manufacture, and the rate of inter- sive gases, Althorrgli iiiosl (if these environments facial processes connected with the materials used; eventually corroclc!cl t tic iriirror metallization, the e.g., interdiffusion and ion exchange. experiment..; wti~~(lil‘i‘i(;iil I to duplicate and the microscopic; ~I~)~)~!;II~~IIII:~:ol’ I lit! specimens was sel- All earth-based solar mirror systems will be sub- dom simil;ir tri II;II iti’;iIly tlt1gr;itled mirrors. Further- jected to terrestrial environmental exposure. The more, t hrtsr! i!xl~~!t~itit(+tiIS (:orit rihuted very little to importani parameters include atmospheric gases devr~1oj)iiig;I 11~i:;if.iitirl~~i.stitiicliiigof the mechanisms (water vapor, oxygen, and air pollutants], UV radia- or kiiii!t ii:s (it I 111, (Ii~~;r;iriiitioti process. tion, temperatures from -3OOC to +15OoC, temperature cycling, mechanical stressing from differentia! ‘I’li(~rt~;ii’ls :;IS{ 181 ,II ~~II!;~II~I:II!(~sources for the different thermal expansion, and dust. Generic problems must t.;i~t,!; of III II.I~III~~II~~!I~~I(I;II ioIi. These include inade- be addressed such as the reactivity [I)of silver with (III;IIO I:II~~I~IIII~:Ill !:I,I:,!, lii~i~irtci mirroring; inadequate

20 Silver/Glass Mirrors for Solar Thermal Systems c:ontrol of the mirroring steps, thus preventing op- cussed Iater in this chapter) and developing a rela- t iinized deposition of silver, copper, or other compo- tionship between the time for performance loss in ricnts; introducing contaminants during the rnirror- accelerated testing and the time for comparable per- ing steps; and reactions of silver and/or copper, formance loss in real time testing (Masterson 1983b). rispecially in the presence of water vapor and oxygen (Masterson 1983a). Clearly, some of these can be c!liminated with rigorous quality control. Preparing Mirrors Using the Wet Chemical Six parameters considered to be primarily responsi- Electroless Process lile for the degradation of a particular type of rnirror The basic steps for preparing mirrors by the chemi- ;is a function of time have been identified (Masterson cal process method consist of selecting and rnanufac- 1983a): (1)humidity (water vapor), [Z) temperature, turing the glass, processing the glass surface, sensit- (:j) thermal cycling, (4)solar ultraviolet radiation, (5) izing the glass surface, depositing the silver reflective environmental pollutants, and (6) mechanical force. Including time, a matrix for testing mirrors actually layer, depositing the intervening copper layer, pro- consists of seven parameters. Each of these has a cessing the copper surface, and backing the metallic surface, usually with a paint. great range of magnitude and gradation, so there is iilmost an infinite number of possible tests to iden- The “wet” or “electroless” chemical process used for lify the impacts of the various individual and combi- metalizing silver mirrors consists basically of nations of parameters. A matrix of finite size was oxidation-reduction reactions. For example, silver is identified for conducting actual experiments to assess initially dissolved in the plating solution as an I he importance of the various parameters (Master- ammonia complex Ag(NH,);. This complex reacts son 1983; Chapter 4, this document). with inverted sugar in basic solutions to plate silver metal onto a substrate immersed in the solution. The In addition to the parameters to which mirror sys- words “wet” and “electroless” are frequently used to tems are exposed, there are several choices that describe the chemical-process mirror because the influence durability. These choices are in five major glass surface is kept wet from polishing to painting topical areas: (1) glass selected, (2) cleaning and and no electric current is used that could cause depo- sensitization of glass, (31 method of depositing silver, sition, as is usually the case for plating metals from (4) backing metal for the silver (if any), and (5) solutions. reflector backing material (e.g., paint or sputtered quartz). Combining optimum methods and materials The chemical components that are used to sensitize that these choices offer for testing the durability of glass, to deposit silver and copper, and that are in a one candidate mirror system could easily involve the commercially used paint are shown in Figure 3-1, preparation and testing of more than 77,000 mirrors along with typical thicknesses of the various layers. (Masterson 1983a). The covalently bonded superstrate is in contact with the metallically bonded silver and copper layers, and However, because of the high performance require- these are bonded to the paint-backing layer. This ments (see Chapter 21,the number of candidate com- section first considers the physics and chemistry binations are not as great as it might seem. In addi- involved in the step-by-step preparation of a mirror, tion, recent work on accelerated testing (see Chapter then discusses the degradation reactions that occur 4) has greatly reduced both the number of degrada- or might occur at the various interfaces, and finally tive parameters and the ranges of magnitude for discusses the properties of both freshly prepared and study. Future research on silverlglass mirrors could degraded mirrors. Some recent research results will focus on elucidating degradation mechanisms (dis- be appropriately included in each subsection,

odified alkyd., lead (not an oxide) IO,, Bone black, Resin, Solvent

uS04.SH20, H,SO4 ydride salt, Iron powder

Sulfur corn pound

Figure 3-1 Morphology and Chemical Components Used in a Typical Commercial Mirror (Thomas 1983) [Not to scale]

Physics and Chemistry of Silver/Glass Mirrors 21 Mirror Preparation Hench (1982) reiterated the danger of assuming that glass is homogeneous. At times, phase separation, Glass. The most economic method for preparing a compositional segregation, devitrification, bubbles, high quality glass with a smooth, even surface is to or unreacted particles may result in accelerated float the molten glass on liquid tin with which it is degradative attack at interfaces. Surface abrasion, insoluble and unreactive. Economy results because polishing relics, scratches, and pits can also cause the glass is so uniform that the normal grinding and accelerated attack (Sanders 1973). Concerning the polishing steps can be eliminated from the manufac- penetration of tin into glass from the float-side, turing process. The glass manufactured by this pro- Colombin (19771 used ion etching and XPS to con- cess, which has an “air surface” and a “tinned sur- clude that the tin concentration varies more rapidly face,” is called “float” glass. In 1958 a general in the first 100 nm than previously thought. In par- composition of float glass was formulated, manufac- ticular, a sharp tin concentration gradient occurs in tured, and marketed (Schweig 1973).The glass com- the first 1.0nm. position is typically SiOz, 72%; Na,O, 14.3%; CaO, 8.2%; MgO, 3.5%; A1,0,, 1.3%; K,O, 0.3%; SOz,0.3%; Silver-to-glass bonding. Several types of silver-to- and Fe,O,, 0.1 wt %, which is the soda-lime silicate glass bonding are possible, including ionic and co- glass shown in Tables 2-2and 2-3 (Chapter 2). All of valent. Only the latter permits an adherent metallic the major glass manufacturers have licensed the layer of silver to be deposited, float glass process from the patent holder, though each may have a-slightly different composition of float glass constituents. Sensitizing. After processing the glass surface to prepare an impurity-free interface, the surface is The next processing step for glass can vary widely sprayed with a solution of stannous chloride di- among mirror manufacturers, In every case, the air hydrate, hydrochloric acid, and, in some cases, wet- surface of the glass is cleaned prior to sensitization; ting agents. The wetting agents assure more com- but in some cases the cleaning steps are numerous plete wetting of the glass surface with the silvering and the cleaning agents may be vastly different. solution. Cleaning is necessary not only to remove dirt and grease but also to remove the hydroxylated silica gel According to Schweig (19733, “Stannous chloride is layer on the glass surface. Mirrors made from freshly necessary for a rapidly deposited, strongly adherent, polished surfaces are known to have better metallic uniform film of silver.” Though other more expensive adhesion than those for which the cleaned glass is sensitizers such as PdC1, have been found and many allowed to stand (Bomar 19811. The best explanation proprietary additives are included in the sensitizing for this difference seems to be that the glass reacts solutions to improve the speed and uniformity of with water to form an alkali-rich layer over the silver deposition, the excellent film adherence is due silica-rich layer (Hench 1982); the layer formed is to the action of the stannous chloride. Stannous chlo- presumably a hydrated alkali carbonate [Hench 1982) ride can also be used to sensitize glass for plating Cu, or a hydrated alkaline earth carbonate (Thomas Ni, or Co mirrors (Saranov 1968). 19833. Recent studies (Pederson 1982 and Thomas 1983) Polishing powders consist of metal compounds, usu- have shown that upon thorough and prolonged rins- ally metal oxides or proprietary mixtures of some or ing of the glass surface after sensitization, all traces all of the following: oxides, hydroxides, carbonates, of chloride ion can be removed. Thus, the tin ion in nitrates, and silicates. For many decades iron oxide, some active form is deposited onto the glass surface or a mixture of lead and zinc oxides, was the powder and its interaction with the silver plating solution used in the polishing slurry. Today, the best polish- components causes the production of the interfacial ing material is cerium oxide, CeO,. The next best adhesion layer which then acts as a nucleus for the compound is zirconium oxide ZrO,, which does not deposition of the reflective silver layer. have the polishing performance of GOz,but is less expensive. Although each manufacturer has its own Although stannous chloride has been used as a sen- proprietary polishing process, the industry standard sitizer for more than a century, it is the least under- polishing material is “powder pack” (Lind 19791. stood step of chemical-process mirrors. Schweig (1973) provided the solution reaction As implied above, experience has shown the manufacturer that the adhesion of the silver film and its optical properties depend upon the cleanliness of the glass surface before silvering. Classically, after pol- SnC1, + H,O = Sn(0H)Cl + HCl ishing, a slurry of powdered chalk, mostly calcium carbonate and some ammonia, is used to clean the surface, though some manufacturers use inorganic detergents, In either case, all traces of the cleaning and suggested t hilt t lit! st!iisif izer adsorbs onto glass compound must be removed before sensitization. to serve as ii ii(icl(!iition silt! for forming colloidal Some manufacturers polish and clean the glass sur- silver part i r:l os, face at the same time by mixing zirconium or cerium oxides into the chalk slurries in proprietary mixtures. Evidenc:ci t hi11 t iri ~~I~li~ri~l~!Imnds to glass is shown in

22 Silver/Glass Mirrors for Solar Thermal Systems the XPS spectrum in Figure 3-2 (Lind 1979).Although In later efforts,Thomas (1983)explained the sensiti- Schweig (1973) did not suggest a bonding mecha- zation differently: The solid manufactured as stan- nism to the glass, he noted that oxidation occurs on nous chloride is actually SnCl2-2H,O, which is only standing; i.e., slightly soluble in water. The real sensitizer, [SnCl,]., is formed in an HCl solution,

3 SnC1, + H,O + 1/2 0, = SnC1, -I- 2 Sn(0H)CI (2) SnC1,*2H2O + HCl - H' + [SnCl,] + 2H,O (7)

and that only fresh stannous chloride solutions The mechanism by which [SnCl,]- bonds to glass and should be used. then bonds silver to glass entails many steps involving insertion reactions, redox reactions, and detailed ligand exchange reactions. This proposed mechanism and all supporting references is discussed in detail in Thomas (1984). Upon sensitization of glass, it is found that (I)tin has bonded to the surface and cannot be rinsed away (Pederson 1980,1982;Schweig 1973; deMinjer 1973), (2) the ability of tin to bond to glass is affected by glass cleanliness and wettability (Daniel 1980; Mills 1980; Goggin 19823, and (3) the chloride ions disap- pear with vigorous rinsing (Pederson 1980, 1982; deMinjer 1973).Thomas' (1983) model for sensitization suggests that [SnCl,]- initially inserts itself between the surface silicon to hydroxide bonds to form a five-coordinate tin complex on the glass Electron Energy, eV surface

Figure34 ESCA Survey of the Unsputtered Air Surface of the Soda-Lime Silicate Float Glass (Pederson and Thomas 1980)

Work by deMinjer (1973) and Pederson (1982) attempted to explain the sensitization process by suggesting the sensitizer adsorbs by the reactions Upon formatian of the tin complex, ligand exchange should be favored; in this way the chlorides may exchange readily with water or hydroxyl ligands during rinsing as illustrated in equation (9).

(9) HoLsi-C]-c1 Cl + 3 H,O - [01?-0H]HO OH + 3 HC1

(5) I OH 0 -Si- -Si- \/ 4\ Si, + SnCl,= Si Sn + 2H' t 3C1- .I / \/ OH 0 and All the reported observations in the literature can be I I (6) explained by reactions (7) through (9) and are sup- ,-"'-OH -Si-0 ported by the XPS and AES data of Thomas (19831 + SnCl,= 0 'Sn + 2H' + 3C1- and Pederson (1980, 1982) as well as by the radio- -Si-0 / tracer data of deMinjer (1973). A shift of 2.8 eV from OlSi-oHI I SnO in the XPS peaks of the Sn 3d,,, and 3d,,, doublet was reported by Thomas (19831, whereas only a 1.7 Similar reactions were suggested for SnC1, adsorp- eV shift is expected from the ligands attached as tion, but with two additional hydroxyl groups at- given in equations (4) through (6). Thus, XPS data Iiiched to each adsorbed tin species. According to provide significant support for the number of ligands 'l'homas (1983), SnCl, is, in reality, a chloride- attached to Sn, as shown by equations (8)and (9).In Ilridged solid (SnCl,], that reacts with water (Cotton using XPS to analyze Si-Sn alloys, Holm (1976) i 1972). reported a chemical shift for the Sn 3d lines of about

Physics and Chemistry of Silver/Glass Mirrors 23 1.6 eV between Sn metal and SnO,. Pederson (1980, that silver does not precipitate as a metal oxide 1982) also reported differences of 1-6 to 1.8 eV for instead of as a metallic film. Finally, the formalde- both Sn" and Sn+-'relative to SnO, In these cases, no hyde is necessary for rapid deposition of the layer additional ligands are attached to increase the mag- within the time span of the actual spraying. The nitude oft he chemical shift to the 2.8 eV measured by silvering step is the most complex since the reducing Thomas (1983). solution, the caustic soda solution, and the complexed silver salt solution must be sprayed on from The Sn complex is restricted to the surface mono- separate spray outlets and must not contact each Ir-lyc?rand only partially covers the available Si sites. other until they are mixed on top of the glass surface. The fractional coverage of Sn complexes on glass In some cases, the silvering step is repeated to surfaces is estimated to be 0.2 (Thomas 19831, 0.1 to increase the thickness of the deposited film. A silver 0.5 (Pederson 1982; deMinjer 1973), and 0.17 k 0.03 layer 70 nm thick is typically deposited in about one (Pitts 4984a). Low coverages are to be expected minute, so trapping impurities at the interface and in bccause of lat era1 electrostatic repulsions in the sur- the film is a distinct possibility. fa(: e in o n o 1a y er. Before discussing the chemistry of silver plating, the In other work, Feldstein (1972) reported improved acid-base chemistry of the glass should be considered. effectiveness of the sensitizer by controlling its con- Penetration of silver into the glass substrate has cent ration. Jones (1972) found that repeated expo- been reported previously (Bastasz 1980; Buckwalter sure and washing improved the effectiveness of 1980; Pederson 1980 and 1982). When the sensitizer PtICll as well as SnCl, - 2 H,O before silver deposi- is sprayed onto the glass, the acidic solution proba- t ion. Buckwalter (1981)reported that contacting the bly protonates most of the surface silicate sites by glass with 1-1 solution of lanthanide rare earths, in exchanging H"s for Na', I<+, CaZ+,and Mg". When the addit ion to tin or palladium chloride sensitizer solu- silver solutions are applied, the highly basic solu- t ions, increased the resistance to delamination of the tions neutralize the acid and thus the silicate sites silver froin glass in the presence of water. I

12Ag(NH,,) ,OH + GNaOH + CliHI20(,- I 0 Ag -Si - 0 - Sriikg Ag, Sr 'Sn/ 12Ag -1- 6HCOONa + 12H,O + 24NH,{ (10) 1 'OH '0' 'Ag, 2Ag(NH,)OH + NaOH + H,CO - and/or 2~g+ HC~ON~ + ~NH,OH + ~NH,~(11) or other possible reactions (Schweig 1973). Besides the silver nitrate salt and sugar components, other components make up the commercial spray silvering solution including formaldehyde, sodium after undergoing an oxicliit ioii-twluct ion reaction, hydroxide, ammonia, and a proprietary additive called a hardener that slows the deposition arid increases the brilliance, density, strength, and uniformity of the reflective layer. In most cases the hardener is a thiosulphate salt. The other three additives are generally needed for the rapid deposition necessary for uniform films using a spray deposition of silvering components. Since silver salts are not soluble in highly basic: solti- tions, the silver ions must be complexed to t'nsitri,

24 Silver/Glass Mirrors for Solar Thermal Systems work of deMinjer (1973) who showed that Pd could trapped in the first few atomic layers of the glass, be bonded to the tin after the silvering step without depending upon the extent of porous “gel” layer for- any loss of silver from the surface. This result clearly mation on glass before plating (Goggin 1981).. suggests that after silver is bonded to the glass, tin is still separately bonded to thc glass, In recent work, The XPS spectrum of a silver surface after spraying, Pitts (1984b) confirmed that both Sn and Ag can be rinsing, and air drying shows the typical impurities detected using XPS and ISS. The Ag signal is three C, 0, S, and Cl from atmospheric exposure, but upon limes that for Sn, which is not possible by any of the etching -30 nm into the bulk, no significant levels of honding hypotheses shown in equation (12). impurities could be found (spectrum 1, Figure 3-3). When the etching process is continued until reaching ‘Thomas’ (1983) model of this process suggests that the silveriglass interface (-90 nm), evidence of Sn Ag’, the only Lewis acid in the plating solution, trapped at the interface was discovered (spectrum 2, attacks the Sn-to-Si bond shown in equation (g),and Figure 3-3) that is consistent with the discussion in leaves behind a covalent Ag-to-Si bond [equation the preceding paragraph. In addition, dramatic in- (1411, which is also consistent with the Auger para- creases in hydroxyl and carbon compounds were meter data of Pederson (1982). observed with SIMS (Thomas 19831. Components of the glass, such as iron, oxygen, calcium, and carbon, Before Ag (NH,), addition, are evident in XPS and SIMS spectra. A contarnina- tion layer at the Agiglass interface has been previously observed, and deuterium oxide experiments (Pederson 19801 suggest that this layer may extend 50 nm into the glass. Shelby (1981) compared the crystallite sizes of both vapor-deposited and chemical-process silver films on silica. The latter were smaller, as deposited, After Ag (NH3)2addition, whereas the former exhibited preferred orientations and were more uniform in thickness. The different durabilities observed for vapor- and chemically dep- OH osited mirrors led some researchers to suspect that +O 2NH3 differences in crystallite size and morphology con- ceivably affect the relative reactivity of the films, Numerous investigations continue about the detailed 0 0 composition or optimization of the silvering solution.

Silver bonded to silicon has been detected using sur- lace SEXAFS as the probe for silver that was vacuum 3.61 cfeposited onto Si(ll1)single crystals (Stohr 1982).It is on this convalently bonded silver that the plating reaction would nucleate. Thomas’ (1983) model suggests that upon reductive formation of the Ag-to-Si Imnd, the Sn complex becomes neutralized and at- caches itself to an adjacent site which it previously could not occupy because of its electrostatic charge. Since the tin complex rebonds to the surface, the possible number of Ag-to-Si bonds is reduced by the number of sites occupied by the Sn. It has been ohserved (Daniel 1980) that the Ag-to-glass bond 1)ttcornes stronger with time. This suggests that the I in may still sensitize Ag-to-Si bonds after forming the metallic Ag layer, possibly by reacting with Ag’ Iionded as surface silicates or trapped as impurities,

Once nucleation has taken place and an electron con- 0 11I I t 300 500 700 900 1100 t300 cliictive layer is formed, the plating reaction should Electron Energy, eV tilke place rapidly, Since the plating reaction is c!ssentially electrode plating after the first surface Iilyer has formed, impurities trapped in the bulk Figure 3-3 XPS Survey Scan of the Elements Present in the Bulk of should not be as concentrated as impurities trapped the Silver Layer (spectrum 1). XPS Survey Scan of the iiI the Agiglass interface. There could be a fairly Elements Present in the Silver/Glass Interface Region Iiirge amount of plating solution constituents such as (spectrum 2). The asterisk indicates Auger signal ll,,O, OH-, Na’, NH;, Sn complexes, etc., that are (Thomas 1983)

Physics and Chemistry of Silver/Glass Mirrors 25 In addition to those mentioned by Schweig (1973), After depositing a copper layer, which is usually one recent patents or papers on improved silvering solu- third to one half the thickness of the silver layer, tions include those by Filip (1969), Sivertz (1972), strong jets of distilled water are needed to disperse Ramanathan (1972),Morimoto (1974),Vaskelis (2975, the coppering solution from the copper surface and 19763, Bahls (1976),Hepburn (1980),and Cottingham especially to remove all traces of the metal powder. (1980). Karasek (1966) developed a solution t.0 im- After rinsing, the mirrors are usually dried on heated prove the uniformity of the silver layer. tables or by warming under infrared heaters before painting the copper surface. Although the next processing step is coppering at room temperature, it is worth noting that any silver film on glass will agglomerate if heated to 100-15OoC At this point the ambient laboratory or household (Czanderna 1965; Lind 1979).The onset temperature atmosphere has little effect on the copper-protected for agglomeration is slightly lower in vacuum than in reflective layer. The copper-backed mirror thus ex- oxygen (Czanderna 1965). hibits a reasonable lifetime for indoor use. If a more extreme environment is encountered, such as a very Backing with copper. There are three methods of humid bathroom, then more protection is needed, In applying the copper backing layer, Electrolytic depo- these cases, the copper layer is protected by a kinetic sition and electroless pouring have almost been barrier such as a layer of shellac or paint. These totally displaced by the spray techniques developed dense polymeric layers slow down the process of in the 1950s. The spray technique for copper deposi- mirror degradation by limiting the pathways that the tion is also an electroless process and thus resembles atmospheric reactants can take to reach the metallic the silver deposition step. The reducing solutions are layers, thus decreasing the exposed surface area. totally different because of the desired rapid deposition in the spraying processes. To deposit copper Elements detected on the copper surface after strip- rapidly, metal powders such as zinc or iron must be ping the paint include Fe and s;some authors (Peder- mixed into the coppering solutions since the electro- son 1980; Mills 1980) have thought that the paint is chemical potentials are great enough for these metals the source of these impurities. However, Thomas to cause reduction. In some cases, extremely strong (1983) prepared a CulAgiglass mirror stack up to reducers such as calcium hydride are used to deposit application of the paint step (Figure 3-4a); an AES the copper onto the silver film. analysis of the Cu surface [Figure 3-4b) showed that The only other chemicals added to the coppering Fe and S were already present, as well as Ca, Na, C, solutions are acids such as sulfuric acid, since metal C1, and 0, without contact to paint or primer. The powders react faster in acidic solutions. In some pro- likely source of sulfur is the plating solution (CuSO, cesses, organic acids such as citric acid are added to and H,SO,). Powdered iron sprayed with the reducer the coppering solutions to improve the smoothness, solution is the logical source of iron; calcium hydride, evenness, and brilliance of the deposited copper the reduction accelerator, is the probable source of layer. Pragmatically, an organic acid additive is not calcium. Sodium, chlorine, and carbon impurities are necessary since copper is used mostly as the backing normally found on samples that are customarily pro- layer and not as the reflective layer. A summary of cessed. Analysis of multiple AES depth profiles into the chemical compositions of various coppering solu- the copper layer indicate that the surface is com- tions is also given by Schweig (1973). posed mostly of oxide compounds.

AES

I Removed Lacquer

I I II I 01 I I I 1 ,,_ 2 4 6 8 10 12 14 16 1 100 300 500 700 v, Minutes Energy, eV

Figure 3-4a Auger Electron Depth Profile Through the Copper Figure 3-4b Elements Present on Copper Surface (Zero Time). Layer of a One Year Old Mirror after Removal of a Shown Uslng an AES Scan (Thomas 1983) Lacquer Backing with Acetone. The profile has been corrected for elemental sensitivity factors as provided by Physical Electronics (Thomas 1983)

26 Silver/Glass Mirrors for Solar Thermal Systems To learn more about the penetration of the oxide The chemical composition of commercially available impurity-containing layer into the copper bulk with paints is a closely guarded secret; the constituents lime, a mirror was made by replacing the paint with a listed in Figure 3-1 were provided as a courtesy of the lacquer to prevent any possibility of reactions with Peacock Laboratories. Most paints consist of a thin- I he inorganic components of commercial paint ner, vehicle, pigment, and drier. Dake (1981) pub- (Thomas 1983). The mirror remained in a laboratory lished a list of 25 commercially available back- environment for one year. After removing the lac- coating and sealant materials. Some representative quer with acetone, the sample was depth-profiled paints used to back glass/Ag/Cu mirrors documented iising AES; the profile of Fe, 0, Cu and Ag each is in some typical patents include a mercaptoalkyl shown in Figure 3-4b. In particular, the iron and alkoxy Si compound (Viventi 19661, silicones (Gen- oxygen impurities persist through the first one third nari 19671, a resin containing bitumen and an epoxy 01' the copper layer, suggesting that a combination of resin (Makijima 19743, a moisture-resistant enamel iron and copper oxides are formed well into the (Nowotniak 19753, polymethyl-H-silozane in CH:,CO c:opper film. Thus, the painticopper oxide interface (Minar 19751, polymeric quinoxaline derivatives iilso consists of a paintiiron oxide interface. (Sinclair 1975), universal paints (Gulf and Western 1976), an epoxy-alurnina-polysulfide rubber-based Ag/Cu interdiffusion. The only direct evidence for a protective coating (Beinarovich 1978), but adiene (:opperdiffusion through silver at 20°C was reported (Sakamoto 1979), acrylic co-polymers (Watanabe l)y Thomas (1983) for a one-year old mirror. Peder- 1979), and benzotriazole (Furukawa 1981). son (1980) observed apparent interdiffusion, but the According to Schweig (1973), "Even the best paint is I opography and morphology of his films caused rtoiibt as to the certainty of interdiffusion. Nearly all not considered good or durable enough for mirrors Ag/Cu interdiffusion studies have been conducted which are. . . used outside. . . . Impermeable and iil)ove 45OoC (Butrymonicz 1974). impenetrable backings are applied in these cases." One of three sheet materials is generally used for 'l*lie only studies on thin films of silver and copper added protection, Thin sheets of metal such as lead, I 11i1I are close to room temperature were from 18G0c tin, or aluminum can be applied over the paint. Very to 220°C (I

Paint layer. After coppering, the nearly complete mir- I.III* is washed thoroughly with distilled or deionized ti,iitcr to remove all traces of solution and metal I)owtler. Gjostein 11973) concluded from AES studies Energy, eV 01 tiictal surfaces that good paint adhesion depends II1I1IIIt I t.il ically on surface cleanliness, particularly with 3 400 500 600 700 800 900 1000 1100 1200 I i*j:;ird to residual hydrocarbon layers. The mirrors Electron Energy, eV

t~t~~btlried, thus terminating the wet processing steps, liiict heated from the uncoated side with infrared to !ritiiperatures of typically 43OC to 65OC. The copper Figure 3-5 XPS Survey Scan of Interfacial Residue on Mirror Paint ~4iii~l'iiceis then coated with a shellac or paints; all After Adhesive Failure. Inset is an AES of paint/copper I oiiiinercial mirrors are protected with paint. surface (Thomas 1983)

Physics and Chemistry of Silver/Glass Mirrors 27 influence of these elements on mirror durability will Unfortunately, the paints actually used, which have be discussed in the next section. been formulated for other applications (Bornar 19811, are especially porous to atmospheric gases, especially water vapor and oxygen. It is not surprising Degradation Phenomena/Mechanisms that Pb, Cu, and 0 have been detected at the copper/ paint interface, but Fe, Ca, C1, and S have also been The physical properties of mirrors prepared by the identified (Thomas 1983; Pederson 1980). In field- chemical process were discussed in Chapter 2. The degraded mirrors, the copper layer corrodes or actu- high reflectance (Table 2-5) can be improved by ally disappears and precedes silver corrosion (Daniel using different glasses (Figures 2-13 through 2-15). 1980: Pederson 1980; Mills 1980). Copious amounts The physical properties of the glass, also presented of chloride are also present. Based on this informa- in Chapter 2, may be properly chosen, but may tion, Thomas (198-) wrote the following reactions change with time. Thus, the chemical and environ- that can occur at the copper/paint interface with mental properties (Table 2-11 are crucial to the life- accumulated water and conducting paths for electrons time of the completed mirror. Performance losses, some of which were mentioned in this chapter, result I I1 from reactions at the silver/glass interface; interdif- H,,O fusion processes at the interfaces; permeation of the PbCl, + 2e-aPb+ Z C1- (-0.268 V) (15) paint followed by corrosion of the metallic layers, particularly in the presence of water and halides; and and from the copper layer, trapped iron filings, and agglomeration of silver, or pinholes left in the films CaH,, there are these following reactions during preparation. Other potential degradation may result from weathering and solarization of the glass, in segregation of glass components that affect a change H,O in the physical properties, UV degradation of paints, Cu-Cu" + 2e- (-0.337 V) (16) and mechanical abrasion of the paint.

In this section, reactions at interfaces and interdiffusion phenomena are discussed. Most of the understanding of the degradation mechanisms has resulted from reports or publications by Czanderna (1978), H l"0 Lind (1979), Bastasz (1980), Buckwalter (19801, 0, + 2H,O + 4e-+40H- (+0.401v) (18) Burolla (19801, Daniel (1980),Mills (1980),Pederson (1980, l982),Shelby (1980),Vitko (1982),Pitts (1984a, $0 1984b),and Thomas (1983).In all the preceding pap- + 2H,Q + 2Cu _I)_ 2Cu (OH)2(+0.77 V) ers as well as that by Bomar (1981), water is identi- 0, (19) fied as an essential agent for causing chemical degradation of the mirrors, but it is also indicated by some that other chemicals participate in, or at least Cut+ -t e- HrO,-+Cut (+0.253V) (20) catalyze, the reactions. The various methods of surface analysis (Czanderna HYO 1975) applied to solar energy materials [Czanderna Cu" + C1- + e--CuCl (+0.538V) (21) 1981) have been extremely helpful for providing the in data required to formulate the chemical degradative H,O mechanisms proposed below. All of these reactions 2H- H2 + 2e- (+2.25 V) (22) ultimately result in a decrease in the solar-weighted - reflectance and thus contribute to increasing the life- cycle costs of solar thermal reflectors. Fe HLFe'+ + 2e' (+0.44 V) (23) Copper/paint. Most paints are subject to UV photo- degradation (Schissel 1980). Nevertheless, they are used to help preserve the mirror. The additives in .commercially available paints are generally used to increase the density of the paint and thus retard the flow of atmospheric reactants to the metallic layers, An ideal paint would be very dense and strongly bonded to the metallic: surface. Diffusion to the metallic layers before corrosion occurs would be time-consuming; the paint density would also retard the outward diffusion of corrosion products from the reaction area and further slow down corrosion react ions.

28 Silver/Glass Mirrors for Solar Thermal Systems 'I'he energy evolved from some of these half reactions detect chlorine in the paint (bulk)probably resulted is sufficient to break copper/paint adhesive bonds from leaching during exposure, which would concen- iind produce corrosive reactions at the silveriglass trate chlorine at the copper/paint interface. Other interface (to be discussed later in this section]. Chlo- authors have also reported that chlorine was present ricle ions, which might accumulate from leaching of in degraded or corroded commercial mirrors (Peder- l)(lC12,aerosol cont arninants or entrained sensitizer, son 1980; Mills 1980). lit~cwell-known depassivators of protective oxides. 'I'he limited amount of surface analytical data obtain- Corrosion of silver and copper. The oxidation of iil)le from the highly insulating paint layers suggests copper and silver in various environments is well I hiit chloride from the paint may be one of the major known and included in many publications (Leid- Ijiwblems in accelerating the corrosion of commercial heiser 1974;Butts 1967),so the related chemical reac- iliirrors (Pederson 1980; Thomas 1983). The XPS tions will not be restated. For solar mirror applica- slwctra in Figure 3-5, taken of the residues left on the tions, some procedure must be devised to exclude ]);lintafter corrosion caused de-adhesion of the paint, corrosive gases from reaching these two metals. The 1)rovide only average elemental composition over the extensive corrosion detected in heliostat s is well whole sample. The spectra show copious amounts of documented (Lind 1979; Daniel 1980, Mills 1980; twth lead and chlorine at this interface; AES analysis Pederson 1980; Bornar 1981). 01' striations left after edge attack of the metallic Vitko [ 1982), using vapor-deposited silver on float Iilyer have shown them to be mostly areas of insolu- glass or vitreous silica and chemical-process mirrors I~lesilver and copper (I) chlorides. These deposits stripped of paint and copper, found that pinholes iiiiiy result from the precipitation of these slightly formed in the silver films when immersed in boiling soluble compounds at the solid/liquid/air interface water (pH = 7.0), as long as the water contained it1 ii way similar to the formation of stalactites and oxygen. The overall dissolution reaction was given s t H lag mi t es. as

Alf'er corrosion caused delamination, AES depth 4Ag + OL + 4H' = 4Ag' + ZH,O (28) j)rofiles into the paint from the Cu/paint interface Iiiive shown great amounts of chlorine on the paint SEM photographs showed that the pinholes enlarge surface (inset Figure 3-51, but almost none could be as a function of exposure (Figure 3-6) and that the i'ound in the bulk of the paint. If lead chloride was a topographical features are similar to those for cor- c:onstituent of the paint, Thomas' (1983) inability to roded heliostat mirrors (Figure 3-7). In accelerated

Figure 3-6 Scanning Electron Micrographs of Silver Films on Float Figure 3-7 Scanning Electron Micrograph of a Field-Corroded Glass Exposed to 1000 rnl of Distilled Water at 90°C Mirror (Vitko 1982) (Vitko 1982)

Physics and Chemistry of Silver/Glass Mirrors 29 testing of complete mirrors, Coyle (1982) used moist Silver/glass interface. Corrosive degradation at the HCl and H,S vapors to evaluate durability. Failure silver/glass interface will obviously result in a loss resulted in 3 to 90 hours in moist HC1 and in 200 to in reflectance for second-surface mirrors. Thomas 1200 hours in moist H,S, depending on the manufac- (1983) identified Si, Ag, 0,Na, I<, Sn, and C1 at the turer. Daniel (1980) found that degradation of silver silver/glass interface. There are three primary types occurs first where the copper backing has disap- of corrosion modes at this interface: (1) reactions peared, presumably by an ion-forming reaction in that result from penetration of reactants through the which copper ions dissolve into a surrounding aque- copper and silver metallization, (2)reactions at the ous medium or the paint. Pederson (1980) also indi- interface from trapped impurities at or near the inter- cated that copper corrosion precedes that of silver. face, and (3) exchange reactions between silver and Mills (1980) interpreted her data as silver diffusion glass with diffusion of silver ions into the glass. to the copper surface, but the exposure of silver might also have resulied from copper corrosion Penetration through the metallization. Several obser- preceding that of silver. The corrosion products on vations have been made that have led the observers degraded silver mirrors include Ci, S, C, and 0 to conclude that reactants penetrate to the silver/ (Czanderna 1978),0,C1, Na, Ca, C, Sn, and S (Peder- glass interface through the silver, copper, and paint son 19801, and C, N,C1, S, and 0 (Mills 1980). layers. Dark nuclei, as observed visually, have been observed in heliostats deployed at Odeillo, France, Sharrna (1978) measured the kinetics of corrosion for Georgia Tech University, the Jet Propulsion Labora- Cu, Ni, and Ag coupons in an air-conditioned office tory, and at the CTRF facility at Albuquerque, New and non-air-conditioned basement for up to two Mexico (Bomar 1981; Burolla 1480). Burolla (1980) years at 22OC. Silver tarnished linearly with time at 3 concluded that the black spots were pinholes formed to 3.5 nin/munth with S and Cl as major surface in the metallization, as illustrated in Figures 3-6 and constituents (asAgCl and AgzS).The data for copper 3-7, probably by the mechanism proposed by Vitko and nickel obeyed a parabolic growth law, and the (1982).However, dark solid phases are formed at this major constituents detected were 0 and C1. Data for interface and can be seen through the glass by usjng Cu and Ni could be duplicated in only one year with optical microscopy (Gupta 1982). an acceleration factor of 100 by using an environmental test chamber, but a correlation for silver cor- These phenomena are illustrated in Figure 3-8, which rosion could not be made. shows photomicrographs taken through the glass of the same mirror before and after a two-week expo- Rice (1980, 1981) studied the indoor (1980) and sure to degradative parameters. The dark portions, atmospheric (1981) corrosion of copper and silver in developed at the silver/glass interface, apparently which the concentrations of SO2,NO2, NH,,, H,S, SR, cause the reflectance losses. They are not pinholes CH,;SH,Cl,, and HC1 were monitored. The corrosion but appear to be silver (OF silver compound) agglo- products for indoor exposures were complex mix- merates, The Combinations of 8OoC, SO2,NO, NO,, tures of hydroxides, carbonates, sulfates or sulfides, Cl,(trace), HzO, and UV apparently cause more vis- nitrates, and chlorides. Typical corrosion rates for ual degradation, but the measured optical degrada- silver at eight population centers in the United States tion is not statistically significant (Masterson 1983b). ranged from 2.8 to 9.0 ng/cm2-h for a 4380-hour Numerous other visual and optical data are available period. This corresponds to about five atomic layers about corrosion products forming at the silver/glass of product per day. In a typical mirror, the silver interface (Lind 1980; Dake 1982; Masterson 1983a consists of about 230 atomic layers. The atmospheric and 1983b). studies (Rice 1981) were conducted in an environmental test facility in which the amount of SOz,NO2, Using surface analysis techniques to probe corrosion O:, NH,,,H20, HZS, and HC1 were controlled. The rate pits into the metallization, Pederson (1980) found of copper corrosion is sensitive to the relative humid- copious amounts of C1 at the silver/glass interface; it ity, but the silver corrosion rate is not. The copper was concluded that an interfacial reaction occurs corrosion rate is influenced by SOz,H2S, C1, HC1, and after grain boundary diffusion of C1 to the interface. Orj.The rate of silver corrosion depends on H2S,Cl,, Thomas [1983) presented AES depth profiling data HC1, and 0, concentrations. For silver, the rate of to show that copper diffuses through silver to the corrosion decreases in the following order: H,S > glass interface within 6 to 12 months. Thus, the visu- NO, > SO, - 0, > C1, > NH, > HC1. ally observed dark products could be either copper or silver compounds formed at the interface. Franey (1981) showed that silver in an epoxy paste Possible reactions at the silver/glass interface. With reacts rapidly with H2S to yield Ag,S. Khrusch potential reactants that are at the silver/glass inter- (1975) produced Ag,O in oxygen by the photostirnu- face (Sn, C1, 0) or that are available from the near lated oxidation of silver films. The wavelength of the interface region (Ca, Na, K, Fe, Cu, Cl), several reac- photons ranged from 110 to 200 nm at 5 mW/cmz and tions can be proposed, Most importantly, Thomas 0.5 8, so it is not known if similar oxidation will [1984b) hypothesized occur at solar wavelengths (>285 nm) and intensities (50-120 rnW/cm2). SnC1,- + 2e- - Sn -I-3C1- (-0.3 V) (29)

30 Silver/Glass Mirrors for Solar Thermal Systems _- plus T and UV

Figure 3-8 Photomicrographs of Undegraded Mirror (a) and After Two Weeks of Exposure to Cornbinationsof T, UV, Pollutants, and Relative Humidity (Gupta 1982)

Si-Ag + H,O + C1-- Products (-0.3 V) (301 reactions (21) and [23)] with a potential of +0.8 V or greater, could rupture the Ag-Si bond. Si-Ag (Interface) - Si (surface) + Ag (metal) (-0.8 V) (31) It is thought that strong adhesion at the silver/glass interface results primarily from forming Ag-Si bonds, Sn + 30H-+ HSn0,- + H,O + 2e- (0.91 V) (32) but this hypothesis is not proven. Thomas (198-1 emphasizes that a conductor is necessary for electron O2 + ZH,O + 4e- 4 OH- (0.4 V) (33) - transport for half reactions occurring at the Cu/paint [Acid) Si + 2H,O - SiO, + 4H' + 4e- and silver/glass interfaces. The copper and silver (0.86 V) (341 metallizations obviously fulfill this need. Liquid H,O is necessary for ion transport. Some of the above (Ihse]Si + 50H-- Si0,-2 + 3H,O + 4e- reactions could be used to explain the improvement (1.70 V) (35) of adhesion with aging (Daniel 1980; Bomar 1982; Althoughequations (16) through (35)are some of the Goggin 1981). 1)ossible half-reactions, it is essential to note that Thomas (1983) also reported that the glass interface solubilities, corrosion products, and thus e.m.f,'s is modified during the sensitizer application; i.e., a tli!pend on the pH of the liquid water. surface carbonate formed by reaction with carbon Although reactions (16) through (35) all could take dioxide is reduced in concentration by the reaction l)li~(:eif the reactants are present, depending on the 2 H' + CaCO, Ca+*+ H,CO, 1'1 I iind other kinetic factors, only a select few actu- - - Ca" + H,O + CO, 1361 iilly contribute to the observed phenomena such as i'tihanced mirror degradation in aqueous media (Lind In related studies, segregation of iron in the glass to 1!W; Bomar 1981; Vitko 1982), impurities detected the silver/glass interface was also reported (Thomas ( lklerson 1980; Thomas 1983), and debonding (Daniel 1983).The implications of an enhanced iron concen- 111110) at the silver/glass interface. For example, tration in this region are not known, but its presence t iriw king the Ag-Si covalent bond requires about -0.8 may assist the observed cohesive failure in glass V. tfalf reactions, both singular and multiple [e.g., near the silver/glass interface.

Physics and Chemistry of Silver/Glass Mirrors 31 Diffusion of silver into glass. Pederson (1980) pub- diffusion coefficients DNa,DAR, D,,, and EDat tempera- lished XPS depth profiles showing depletion of Ca" tures as low as 100OC for several glasses. and Na*from 30 nm to 80 nm into the glass from the silverjglass interface (Figure 3-9). Evidence for silver The aggregation and migration of ion-implanted ion diffusion into this zone was also presented by silver in a lithia-alumina-silica glass were studied by comparing depth profiles into glass and silver on Arnold (1977) using RBS. Colloidal aggregates were silica. Daniel (1980) claimed that silver penetrates as formed during implantation at room temperature and deep as 500 nm into a glass reaction zone that is also during annealing up to 35OOC. considerably enhanced by reactions with water. Gotz (1972) used IR and microgravimetric analysis Exchange of Ag' with Na" in the glass was also to measure the solubility of water in commercial reported by Buckwalter (1980)and Ag'diffusion into float glass from 120OOC to 147OOC. Gottardi (1972) glass by Mills (1980), measured the kinetics of extracting Na' from float glass with H,O by using a nuclear reaction analysis technique. Concentration profiles for Sn and other components were also obtained. Glass weathering/solarization. Solarization of glasses alters the optical properties by inducing chemical changes of the constituents in glass. Vitko (1980) discussed the magnitude of this effect and found a 2.5% improvement of the transmittance, presumably resulting from the conversion of Fe+' (solar absorb- Y Na I ing) to Fe+3(nonsolar absorbing). Lind (1980) found that only minor changes occurred in the optical properties of glass after 40 years. Weathering will result from compositional changes of the glass by attack from atmospheric constituents, ._0 primarily water vapor. In its simplest form (Higgin- botham 1980), hydrolysis of the surface occurs by 0 '.*\ \ I I

%utter Time, min. =Si-0-Na + H,O - = Si-OH + Na' + OH- (37) 0 400 800 1200 Approximate Depth, A followed by, providing the leached alkali is not removed from the glass,

ESi-O-SiE + OH- -. rSi-OH + rSi-0- (381 Figure 3-9 ESCA Depth Profile of the Surface in Figure 3-2 (Peder- son and Thomas 1980) A more complete discussion of weathering is given by Hench (1982). Bastasz (1980) found limited, if any, silver diffusion A recent bibliography on weathering was assembled into glass, but his films were vacuum-deposited.The by Godron (1976); the text is In French, but the 101 combination of all the above results suggests that references, 20 tables, and several figures and reac- water at or near the interface or trapped impurities tion mechanisms are valuable assistance. An earlier may enhance the Ag' formation or diffusion. One of review by Jelli (1974a) with 26 references is also in the macroscopic consequences of the ion exchange French. Isard ( 1981) compared weathered glass sur- reaction zone into glass is that cohesive failure faces with leached ones and concluded that the sur- occurs through this region. Mills (1980) provided face reactions proceed by different mechanisms. surface analytical data to support the phenomeno- logical observation; i.e., SAM scans showed that Si Hench (1982) emphasized the necessity of using sur- was present on both sides of the silver/glass interface analytical techniques such as AES, SIMS, NRA, and secondary ion photoemission spectros- face after forced delamination. XPS, IR, copy to monitor the changes in glass composition Exchange and diffusion of Ag' and Na" in glass has and to relate these to weathering phenomena. Exam- been studied extensively using molten salts and ples of recent studies using AES on weathered glass temperatures of 25OoC to 6OOOC (Doremus 1968; surfaces are those by Chappell (1974), Gjostein Williams 1975; I

32 Silver/Glass Mirrors for Solar Thermal Systems Finally, Kornppa (1978) found that glass-surface available. A calcium hydride reaction with water durability was greatly improved by storing glass for releases enough energy to cause energetic side reac- a few weeks before heat-treating it at 30Ooc. A hint tions in the glass that could affect silver-silicon that a vertically integrated mirror industry might be bonds at the silver/glass interface. Thus, if the paint desirable could be gleaned from this comment and sealed out everything but water, even at the edges, from those given in Bomar’s (19811 document. degradation due to impurities could still take place as soon as the water diffuses to the metallic layers or Discussion of corrosion in the mirror system. Segre- interface regions. gation of glass components to the silver/glass interface (e.g., iron compounds), together with the forma- Dust. The reflectance of mirrors is adversely affected tion of a gel layer, in time could weaken the glass by surface soiling, and generally the loss in perfor- structure at this interface, In fact, examination of mance increases with the amount of soil accumulated different glasses, silvered by using several different on the glass surface. The rate of dust accumulation methods and subjected to adhesion fracture tests at depends on the site location, the stowed position of SERI and by others (Buckwalter 1980),revealed that heliostats, mirror type, weather, etc. (Cuddihy 1980). cohesive failure of glass occurred instead of adhesive Most importantly, the adhesion of dust to glass is a failure at the silver/glass interface. basic surface chemical reaction that has been studied only obliquely. Cuddihy (1980) indicated that the Segregation of paint components to the copper-paint following criteria appear to be required for mirrors interface may initiate copper degradation, but none and photovoltaic cover plates: hard, smooth, hydro- of the electrochemical reactions that could take place phobic, low-surface energy, and chemically free of within the copper layer should be able to cause deg- sticky materials and water-soluble salts. Low-soiling radation reactions within the silver layer, Since the environments or site locations should have low con- silver layer is also observed to undergo dissolution if centrations of organic vapors, frequent rains, low one assumes a basic pH environment, two possibili- dew and relative humidity, and few dew cycles. The ties are presented. First, chloride from the paint current evolving strategies include cleaning and lowers the potentials needed to cause silver to react, washing; Morris [ 1980) discussed cleaning methods and this reaction will continue as a favored process to test the feasibility of implementing the strategy. until the chlorine is exhausted. Secondly, copper dif- Measured reflectance losses resulting from dust fusion into the silver and then its reaction with oxy- accumulation have been reported for mirrors located gen at the silver surface and grain boundaries causes in a variety of environments by Pettit (1978),Black- unreacted metallic silver to agglomerate. At elevated mon (1978), Morris (1980),and Roth (1980). temperatures, increased oxygen absorption by me- Mechanical stresses. Degradation in performance tallic silver may enhance this degradation. may result from various physical processes that are Thus, under basic pH conditions, the supply of oxy- combined in this paragraph as mechanical stresses. gen and chlorine together, even though the chlorine The strength of glass can be changed by impact of quantity is limited, may account for the total corro- sharp particles (Wiederhorn 1979a and 1979b) or by sion or dissolution of the Ag/Cu metallic layer which water vapor that alters the crack velocity growth occurs in accelerated tests within eight weeks (Mas- from residual stresses in the glass (Zwissler 1981).In terson 1983b). To corrode silver under acidic condi- actual use, stresses will be induced in the glass by tions, only 0, and H20need be present (Vitko 1982). wind loadings on the heliostat, temperature extremes, The paint delays direct attack on the metallic layer and thermal cycling (Bomar 1981). Shrinkage of the but may contribute to enhancing edge attack by pro- paint may also produce stresses, so debonding at the viding C1- ions. The diffusion of copper into the silver silver/glass interface can result from a variety of may be the rate-determining step in silver corrosion mechanical stresses. Generally, the weakest part is under basic pH conditions during accelerated testing. not the zone near the silver/glass interface, but in the glass. The stress sensitivity of the glass/Ag/Cu multi- The impurities observed to be trapped in the copper layer stack is also indicated by attempts to use thick layer, at the copper/paint interface, and at the silver/ copper films which cause delamination at the silver/ glass interface, are factors in reducing mirror life- glass interface (Bomar 1981). time from a goal of 20 years. The variability of impurity levels may be the primary reason some mirrors fail long before others, even when made by the Influence of Degradative Processes on the same manufacturer (Lind 1980; Pohlman 1980; Mas- Properties of Mirrors terson 1983a). For example, water trapped at the silver/glass interface can be converted by electro- Optical reflectance. The most important aspect of a chemical reactions or ion exchange reactions to heliostat is the optical efficiency (see Chapter 2) that hydroxide ions which can then attack the glass. directly affects the energy gain of the central receiver. According to equation (g), reduction of tin chloride Optical losses can quickly limit the cost-effectiveness complexes at this interface will also release hydrox- of a central receiver electrical generation system. ides to attack the glass. Calcium hydride and iron

I trapped in the copper layer can participate in reduc- Optical losses can result from Fe*2absorption in the tion reactions as long as water or counter ions are glass, defects in the glass introduced or excluded

Physics and Chemistry of Silver/Glass Mirrors 33 during processing, debonding, and dust accumula- Mirrors Made Using Vacuum Processes tion. These losses seem to be “designed-in” from economic considerations. For example, low-iron float The preparation, degradation, and influence of deg- glass is less expensive than float glass with no iron radation on the properties of silver/glass mirrors present. The selection of float glass does not permit made using vacuum processes have been studied flexibility for minimizing manufacturing defects, nor only recently, compared with those described pre- does it provide a means of choosing a surface that is viously in this chapter. The preparation method is not reactive with dust. Furthermore, paints that can much less involved, and using vacuum offers better be applied to copper are numerous and inexpensive, opportunities for securing a mirror with a 20-year but almost none can bond to silver with any durabil- lifetime. The basic steps for preparing mirrors by ity. Thus, economics suggest using the inexpensive vacuum process methods consist of selecting the glass and leaving the copper layer intact but min- glass substrate, cleaning and processing its surface, imizing it. silvering the glass, and applying a backing to protect the silver from the environment. Other losses in reflectance result from chemical processes in the multilayer stack. Scattering centers or pinholes develop at the silver/glass interface because Preparation of Mirrors in Vacuum of corrosion. Diffusion of silver into the glass may reduce the thickness to less than about 55 nm, which Glass. The criteria for selecting a glass for solar is the calculated uniform thickness needed to yield a applications are the same as those discussed in 97°/b reflectance for a pure silver film. Interdiffusion Chapter 2. For vacuum processing, however, the of copper and silver may also change the reflectance glass thickness can be chosen to accommodate high- from that of silver to that of a AgiCu “alloy.”Impuri- throughput vacuum technology. Thin glass with ties entrained at the silver/glass and copper/paint minimal losses from Fe” absorption (Figure 2-3 in interfaces can result initially in debonding at the Chapter 2) or Corning 7809 (Coyle 1981b),where iron silver/glass interface, which is a consequence of the is in the nonsolar absorbing ferric state, become reactions, to produce optical losses. Finally, the options that are not as easily adapted to the chemical- entire metallization may be destroyed by corrosive process mirroring equipment. reactions. The glass surface must have the gel layer removed and be cleaned prior to insertion into the vacuum chamber. Because of the stacked makeup of the mirror, corro- Polishing steps similar to those previously described sive attack begins at the paint/copper interface and may be used, followed by using standard cleaning works toward the silver. If the mirror is free from treatments for objects to be placed in a vacuum pinholes, the progress of corrosion is restricted to the (Baouchi 1983).Borisova (1973) used radio frequency gradual movement of chemical reactions through the processes to remove the gel layer. After insertion into metallizations. Only after the silver is corroded so the vacuum chamber, sputter cleaning of the glass that its thickness is less than that needed for good substrate may also be employed (Holland 1956). No reflectance does failure occur. Failure is seemingly systematic study of the influence of glass cleaning on sudden. Thus, many of the mirrors within a heliostat the properties of vacuum-deposited silver mirrors field that had been performing satisfactorily for for solar applications is known to be published. years can fail precipitously within weeks or months When a suitable vacuum has been reached; e.g,, ca. of each other. 0.1 Pa, the silver layer can be deposited by one of several methods. Those methods found in recent literature include sputtering, thermal evaporation, and ion plating. Mechanical strength. Various chemical reactions and physical processes can cause mechanical stresses Silver deposition. Sputtering. Sputtering is a process that ultimately reduce the system performance. Ob- in which the material to be deposited is bombarded viously, stresses that cause failure in the glass are with energetic ions [- 2 to 10 keV). Atoms from the not desired, and these do not seem to be a serious target are ejected and deposited onto a substrate problem. However, the glass near the silver/glass positioned properly for collecting them. Although interface can easily become weakened by the corro- Kienel (3981) did not prepare silver mirrors, he de- sive reactions with the metallization, causing crack- scribes an automated cathode sputtering system that ing not found in pure glasses. The shrinkage of paint can be used for making mirrors 10 m2in size. He also can result in exposing the metallization to premature cites a number of advantages of sputtering over corrosion, or it can cause debonding at the silver/ thermal evaporation processes. Fan (1981) prepared glass interface if adequate stress relief is not pro- glass/TiO,/Ag/TiO, multilayers by an r.f. sputtering vided through the ductile metallization. A potential method. Since his objective was to secure partially problem is the effect of weathering on the hail impact transmitting rnultilayers, the silver layer was only strength of the glass, but little information is 13 nm thick. The same method could be adapted eas- available. ily for making mirrors.

34 Silver/Glass Mirrors for Solar Thermal Systems Hartsough (1977) used planar magnetron sputtering ious substrates, but not on glass. Generally, the at rates above 12 nmis to prepare 100-nrn thick alum- properties were comparable with those of r.f, sput- inum films with reflectances as good as those made tered films. by thermal evaporation. Pitt (1981) used r.f. sputtering to prepare planar optical waveguides on microscope slides, but also prepared samples in which Na4 Degradation of Mirrors Prepared in ions in glass were replaced with Ag' using field- Vacuum assisted diffusion. Thus, an alkali pre-depleted zone could be replaced with Ag' followed by silvering the Several sputter-deposited silver mirrors subjected to glass to make a mirror rather than a waveguide. salt spray and HC1 accelerated degradation provided comparable durability to a chemical-process mirror Adams (1979) prepared co-sputtered films of silver used as a control (Jorgensen1980). The (glass/silver/ and aluminum for reflector applications. The method backing/paint) configurations that met acceptance could be adapted to preparing reflecting films of are: 7809/silver/stainless 304/paint; 7809/silver/ graded composition and desired properties. The re- Inconel/paint; soda lime or 7809/silver/copper/paint; flectances of all the Ag/A1 alloys were less than that soda lime or 7809/silver/Mo/paint; and 7809 silver of pure silver. [chemical process)/copper/paint (control]. Sputter- Carmichael (1975) described an apparatus for the deposited silver mirrors did not survive accelerated continuous production of sputter-coated glass prod- tests nearly as well as the best chemical-process mir- ucts, such as glass sheets. Typically, the glass could rors that were used as a control (Masterson 1983b). be ion-bombarded for cleaning before deposition of There were no accelerated tests of mirrors made by films. Multiple coatings could be applied by using Lind (1982) or Masterson (1983a and 1983b) using vacuum locks between several chambers. thermal evaporation.

Jorgensen (1980) lists 25 mirrors that were pre- No standard-stack ion-plated mirror performed as pared by sputtering 100 nm of silver onto soda-lime well as the standard chemical-process mirrors, but and Corning 7809 glasses. Different backing metals glass/Ag/Cu/Ni (electroless)/paint mirrors showed were used, as well as paint. The mirrors were sub- promise of being more durable than the durability of jected to salt spray and HC1 acid-vapor accelerated the standard chemical-process mirror (Lind 1982). tests. Masterson (1983b) also subjected sputter- deposited silver mirrors to accelerated testing at 80°C in 80% RH, with SO, and NO, pollutants, and Influence of Degradation Processes on the with UV exposures. Reflectance Thermal evaporation. Vacuum evaporation of silver No studies of the mechanisms of degradation have onto a substrate results from heating silver metal in a been conducted on mirrors made by vacuum pro- boat, basket, or other evaporant holder for deposi- cesses. Solar-weighted reflectance is the only prop- tion onto a substrate, The process is simple and erty used to measure the survivability in the three excellent for laboratory work, but not as adaptable to sets of accelerated tests (jorgensen 1980; Lind 1982; production processes as sputtering. Films made by Masterson 1983b).In all cases, reflectance decreases other processes are usually measured against evapo- from its initial value as the si1ver:corrodes in the rated films as the standard. Silver films with 97°/0 environmental exposure or as scattering centers are solar-weighted reflectance have been routinely pre- formed at the silver/glass interface. pared [Call 1980). Hass (1975) used vacuum processes to prepare a Ion plating. With ion plating, the substrate is used as mu1 tilayer stack of subs trate/Al,O,/Ag/Al,O,/SiO,, a target for stopping ions accelerated at a source. The where the substrate glasses were silica, polished material liberated from the source drifts to and coats Cer-Vit, and fire-polished microscope slides. The the surface to be plated. Ion-plated silver mirrors A1,0,, and SiO, thicknesses were 30 nm and 100-200 were prepared in the following manner (Lid 1982): nm, respectively. The original reflectances exceeded Ions were evaporated into an argon-defined r.f. 95%; after passing a transparent tape adhesion test, plasma from a boat near ground potential. The glass the mirrors were tested at and survived a 195-hour substrate was biased at 500-700 V, and multilayer exposure to moist H,S at 20-25OC without a measur- stacks were prepared with typical thicknesses of 100 able loss in reflectance. nm of silver/100 nrn of silver plus another metal and an overcoating metal alone (OMA) where the OMA thicknesses were 400 nm for Cr, 1000 nrn for Ni, 1000 Mirrors Made Using Organometallic nm for 304 stainless steel, 1500 nrn for Al, and 1500 Solutions nm for an A1 (35%)/Cu(65%) alloy. Soda-lime silicate and Corning 0317 glasses were used as substrates, The reaction that deposits silver from an organome- The reflectances were typically 89% to 91%. The mir- tallic solution is basically a thermally induced de- rors were subjected to accelerated testing. Yoshihara structive oxidation-reduction reaction. At elevated (1977) also prepared ion-plated silver films on var- temperatures of 200°C to 7OO0C, depending on the

Physics and Chemistry of SilverlGlass Mirrors 35 proprietary organornetallic solution used, the organic resist at t tick bet ler than some backed vacuum- ligands that complex the silver atoms or ions decom- deposited mirrors. This probably results in part from pose. Since the decomposition products are usually the thickness of the film [up to 700 nm) as well as its volatile at elevated temperatures, they evaporate and strong interfacial bonding. The strong bonding is metallic silver is left as a coating on the heated sub- likely caused by depletion of oxygen from the glass strate. This process is in use but mainly for manu- surface during decomposition of the organometallic facturing filins used as anti-glare, heat-reflecting, or (Pitts 1984b);the reactive surface permits silver to decorative coatings. The organometallic process is forin bonds with silicon from the glass, resulting in a slightly more expensive and gives slightly less re- strong interfacial bonding. flective films than the electroless "wet" process. Accelerated testing of organometallic mirrors made with and without an adhesion promoter by the Mirror Preparation Engelhard Corporation was accomplished by Mas- terson (1983b).Their durability was shorter than for Glass. As in the glass processing cases described the chemical-process mirrors, but it is not known if above, the substrate is cleaned or polished. Advice the durability to accelerated degradation was a for cleaning the substrate is often available from the direct correlation to lifetime for normal use, manufacturer of the proprietary solution. Research topics associated with organometallic mir- Silvering. Numerous manufacturers make organome- rors include reflectance enhancement, minimizing tallic silvering solutions; a recent listing is given by the thickness of silver, and eliminating voids in the Schweig (1973). Although the exact composition is films [Pitts 198411). The strong adhesion and simplic- proprietary to each supplier, the solutions usually ity of processing are attractive potential advantages. contain met a1 resinates or sulforesinates in organic solvents. For further details, a search of the patent literature is required. Langley (1974)published some background information on organonietallic processing and includctd a complete literature review through Mirrors Made Using Other Processes 2065. In this section alternative silvering steps in a mirror For silvpring a r:lecined glass substrate, the solution preparation sequence will be cited. None of these is is sprayed 01%spun onto the siirface by dropping the known to be in current widespread use, nor are any thick paste into the c;entt?r of the rotating surface of durability properties or degradation mechanisms the glass; the angular rotation of the substrate known about them. spretids the solution over the surface, which may or may not be heated at this stage. If it is at room Electroplating silver from solutions is described in tcimpvrature, tho substrate is then heated to initiate numerous textbooks on electrochemistry. It was the decomposition of the solution. Once the reaction replaced by the chemical process primarily for eco- has gone to completion, the substrate is cooled and nomic reasons (Schweig 1973). It is not known if a the remaining steps of the manufacturing process are recently patented process for high speed electro- conipleted 10obtain a finished mirror. deposition will alter the economic factors (Rymwid 1981). No known chemical vapor deposition (CVD) IJsing solutions and mirrors made by Englehard process is known for depositing silver onto substrates Corporation, reflectances of 85% to 93% were mea- (Schweig 1973).CVD processes exist for depositing sured on Corning 7809 and 0371 glasses (Pitts 1984b). Ni, Cr, As, and Al, but these all have lower solar- The silver thicknesses were 300-nni to 600-nm thick, weighted hemispherical reflectances than silver. which is 5-10 times more than is necessary to obtain the maxiinurn reflectance of silver in a thin film. Hamm [ 19771 patented a process in which a mixture of I

36 SilvedGlass Mirrors for Solar Thermal Systems Protective Coatings for Reflectors minum-Silver Alloy Mirrors for Use as Solar Reflectors.” Thin Solid Films, Vol. 63 (No.1); pp. Call (1980) presents a broad overview of reflector 151-154. materials, including protective coatings, and the Arnold, G. W.; Borders, J. A. (April 1977).“Aggrega- application of paisive thin films for solar application and Migration of Ion-Implanted Silver in tions. Several recent papers have been published on Lithia-Alumina-Silica Glass.” Journal of Applied specific coatings that are being studied for backing Physics, Vol. 48 (No. 4); pp. 1488-1496. silver mirrors. These are all from the comprehensive summary of possible coatings (Dake 1981). Bahls, H., inventor. Peacock Laboratories, Inc., assignee. (28 September 1976). “Method for Apply- It seems obvious that the solution to silver corrosion ing Metallic Silver to a Substrate.” Patent 3,983,266. is to exclude residues principally by isolating silver between impervious dielectric layers. Hence, it is no Baouchi, A. W. (1983). Preparation and Properties of surprise that most efforts are directed toward depos- a Multilayer Reflector Stack Consisting of Glass! iting oxides such as Al,O, (Davies 1971; Wille 1971; AI,O,/Al/Ag. M.S. Thesis: Denver, GO: Univer- Adams 1972; Hass 19753, cuprous oxide (Franz 1976; sity of Denver. Breininger 1982),and iron oxide (Franz 1976). Ebert (1982), taking a broader approach, has reactively Bastasz, R. (September 1980). “Auger Analysis of sputtered TiOz,BeO, In,O,, SnO,, and SiO, onto glass Silver-Glass Interfaces.” Solar Reflective Mate- substrates; his methods could be adapted to silver rials; Proceedings of the Second Solar Reflective mirrors, Pearlstein [ 1979) deposited Rh onto silver to Materials Workshop; Sun Francisco, CA; 12-14 isolate it from reactants, Mattox (1980) reviewed February 2980. Appears in Solar Energy Mate- briefly the various vacuum methods available for rials, Vol. 3 (No. 1, 2); pp. 169-176. depositing coatings onto various substrates. Lebert Beinarovich, L. N.; Bakhtiyarova, F. Sh.; Kozhev- (1980) favors sandwiching silver coatings between nikov, Yu G. (1978). Protective Coating for Sil- two glass panels whenever possible. vered Mirrors; U.S.S.R. 550,354. Soviet Invent. Recent papers dealing with organic coatings for 111, VO~.Y (NO.48); pp. 1-2. copper or silver metals and a discussion of their per- Bieg, K. W.; Wischmann, K. B. (September 1980). formance include using benzotriazole on copper “Plasma-Polymerized Organosilanes as Protective (Notoya 1979), plasma-polymerized organosilanes Coatings for Solar Front-Surface Mirrors.” Solar on silver (Bieg 1980), silicone resins on first-surface Reflective Materials; Proceedings of the Second silver (Dennis 1980), and polymer-based ceramics Solar Reflective Materials Workshop; San Fran- for backing mirrors [Borisova 1975). cisco, CA; 12-14 February 1980. Appears in Solar Energy Materials, Vol. 3 (No. 1, 2); pp. 301-316. Adhesion Enhancement Blackmon, J. B.; Curcija, M. (1978).“Heliostat Reflec- tivity Variations Due to Dust Buildup Under Adhesion at all interfaces in a multilayer stack is a Desert Conditions.” Proceedings of the Seminar on natural requirement for durable mirrars. In addition Testing Solar Energy Materials and Systems; Gai- to the above discussion about adhesion of silver to thersburg, MD: 22-24 May 1978. Mt. Prospect, glass (e.g., role of silver-silicon bonds in chemical- IL: Institute of Environmental Science: pp. 169-183. process, vacuum-deposited, and organometallic mirrors), some recent specific studies were identified. Bomar, S.;Keith, D.; Maddox, K.; Walton, J. D. (1982). Brewis (1967) discussed the surface properties asso- Proceedings: Durability of Reflecting Surfaces ciated with adhesion between metals. Larry (1975) Used in Solar Heliostats. Georgia Institute of used NiO to improve silver adhesion onto substrates Technology, Engineering Experiment Station, and extended his work (1976) to include PbO, CaO, Atlanta, GA, 1982. A1,0,, Bz03,SiO,, and Bi203.The intended application of his efforts was toward conducting metalliza- Borisenko, Y. N.; Gritsyna, V. T,; Pershin, V. F. tions rather than preparing solar mirrors. Borisenko (1974). “Effect of Hydrogen(+) Ion Irradiation on (1974) studied the effect of H’ ion bombardment on Adhesion of Silver Films to Glass.” Vzaimodeist- the adhesion of silver films to glass. vie Atom. Chastits 5 Tverd. Telom. (No. Ch. I);pp. 142-144. In Russian. References Borisova, I. I.; Subolevskaya, E. S,; Afonia, N, S.; Moiseeva, V. P. (May/June1973). “Preparation of a Adarns, H. N.; Collingswood, N. J.; inventors. Denton Glass Surface of Vacuum Coating.” Glass Ceram- Vacuum Corporation, assignee. 1.29 August 1972)+ ics, Vol. 30 (No. 5, 6); pp. 317-319. “Protective Coating for Surfacks of Silver anb Mirror Fabrication.” U.S. 3,687,713. Borisova, I. I.; Afonina, N. S; Kiryukhina, N. A.; Loginova, T. V. (19751, “Industrial Methods of Adams, R. 0.;Nordin, C. W.; Fraikor, F. J*; Master- Applying Protective Coatings to Mirrors.” Stekl son, K, D. (15 October 1979). “Sputtering of Alu- Kerarn,, Vol. 10; pp. 9-11. In Russian.

Physics and Chemistry of Silver/Glass Mirrors 37 Brqininger, J. S.; Greenberg, C. B., inventor. PPG Colombin, L; Jelli, A; Riga, J; Pireaux, J. J.; Verbist, J. Industries, Inc., assignee. (9 February 19821. “Direct (1977). “Penetration Depth of Tin in Float Glass.” Electrolysis Decomposition of Cuprous Oxide Films Journal of Non-Crystalline Solids, Vol. 24 (No. 2); on Glass.” U. S. Patent 4,315,055. pp: 253-258. Brewis, Derek M. (January 19671. “Surface Proper- Coombe, R. D.; Wodarczyk, F. J. (1November 1980). ties Associated with Adhesion.” Polymer Engi- “UV Laser-Induced Deposition of Metal Films.” neering and Science, Vol. 7 (No. I); pp. 17-20. Applied Physics Letters, Vol. 37 (No. 9); pp. 846-848. Buckwalter, C. Q.; McVay, G. L. (September 1980). “Inhibiting Degradation of Wet Process Manufac- Cottingham, J. G., inventor. U. S. Department of tured Solar Mirrors.” Solar Reflective Materials; Energy, assignee. (7 October 1980). “Method of Proceedings of the Second Reflective Materials Making Reflecting Film Reflector.” U. S. Patent Workshop; San Francisco, CA; 12-14 February 4,226,657. 1980. Appears in Solar Energy Materials, Voi. 3 Cotton, F. A; Wilkinson, G. (19721. Advanced Inor- (NO.1, 2); pp. 215-224. ganic Chemistry; A Comprehensive Text. New York: Interscience Publishers; 1145 pp. Buckwalter, Jr., C. Q.; inventor. U. S. Department of I Energy, assignee. (25 August 1981). “Process for Coyle, R. T.; Barrett, J. M,; Call, P. J. (October 1981a). Preparing Improved Silvered Glass Mirrors.” U. S, Durability of Silver-Glass Mirrors in Moist Acid Patent 4,285,992. Vapors. SERUTP-641-623. Golden, CO: Solar 1 Energy Research Institute; 41 pp. Burolla, V. P. (September 1980).“Deterioration of the Silver/Glass Interface in Second Surface Solar Coyle, R. T.; Livingston, R. E. (December 1981b).Eva- Mirrors.” Solar Reflective Materials; Proceedings luation and Market Testing of Corning Code 7809 of the Second Solar Reflective Materials Work- Thin Solar Glass. SERI/TR-733-1230, Golden, CO: shop; San Francisco, CA; 22-14 February 2980& Solar Energy Research Institute; 27 pp. Appears in Solar Energy Materials. Vol. 3 (No, 1, 2); pp. 117-126. Coyle, R. T.; Barrett, J. M.; Call, P. J. (March/April 1982).“Durability of Silver-Glass Mirrors in Moist Butrymowicz, D. 3.;Manning, J. R.; Read, M. E. Acid Vapors.’’ Solar Energy Materials, Vol. 6 (No. (19741, “Diffusion in Copper and Copper Alloys; 3); pp. 351-373. Part 11. Copper-Silver and Copper-Gold Systems.” Journal of Phys. Chem. Ref. Data, Vol. 3 (No. 2); Cuddihy, E. F. (September 1980). “Theoretical Con- pp. 527-602. siderations of Soil Retention.’’ Solar Reflective Materials; Proceedings of the Second Solar Reflec- Butts, A.; Coxe, C. D.; editors. (1967).Silver Econom- tive Materials Workshop; San Francisco, CA; 12- ics, Metallusgy, and Use. Princeton, N.J: Van 14 February 1980. Appears in Solar Energy Mate- Nostrand; 488 pp. rials, Vol. 3 (No, 1, 21; pp. 21-33.

Call, P. J. (1980j. “Applica’tions of Passive Thin Czanderna, A. W.; in K. H. Behrndt, editor. (1965). Films.” Polycrystalline Amorphous Thin Film Vacuum Microbalance Techniques; Vol. 5. New Devices. Edited by L. L. Kazmerski. New York: York: Plenum; pp. 135-147. Academic; pp. 257-291. Czanderna, A. W.; editor. (1975).Methods of Surface Carmichael, D. C.; Chambers, D. L.; Wan, C. T.; inven- Analysis. New York: Elsevier Scientific Publish- tors. Shatterproof Glass Corporation, assignee. (9 ing Company; 481 pp. September 1975).“Apparatus for Continuous Pro- Czanderna, A. W. (1978);in B. L. Butler, SERI/TR-31- duction of Sputter-Coated Glass Products.” U.S. 042. SERI Materials Branch Semi-Annual Report, Patent 3,904,506. p. 13.

Chappell, R. A,; Stoddart, C. T. H. [October 1974). Czanderna, A. W. (October 1981).“Stability of Inter- “Auger Electron Spectroscopy Study of Float Glass faces in Solar Energy Materials.” Appears in Solar Surfaces,” Physics and Chemistry of Glasses, Vol. Energy Materials. Vol. 5 (No. 41; pp. 349-377. 15 (NO.5); pp. 130-136. Dake, L. S.; Lind, M. A. (September 1980). Adhesion Colombin, L.; Jelli, A,; Charlier, H,; Caudano, R.; and Chemical Vapor Testing of Second Surface Pireaux, J. J.; Riga, I.; Tenret-Noel, C.; Verbist, J. Silver/Glass Solar Mirrors. PNL-3575. Richland, (1976). “Methods for Surface Studies and Chemi- WA: Pacific Northwest Laboratories; 63 pp. cal Profiling in Glass.” Physics in^ Industry; Pro- ceedings of the International Conference; Dublin, Dake, L. S. (February 1981). Summary of Solar Ireland; 9-23 March 1976. Edited by E. O’Mongain Thermal Optical Materials Durability Testing for and C. P. O’Tolle. New York: Pergarnen Press; pp. FYBO-81. PNL-3756. Richland, WA: Pacific North- 113-116. west Laboratories,; 75pp.

38 Silver/Glass Mirrors for Solar Thermal Systems Dake, L. S.; Lind, M, A. (April 1982).Effect of Expos- Filip, R:; Horale, M.; Biciste, V.; inventors, [15 Janu- ing Two Commercial Mirror Manufacturers Second- ary 1969). Czech. 130,534. Surface Silver/Glass Mirrors to Elevated- Temperature, Mechanical Loading, and High Hu- Franey, J. P.; Graedel, T. E.; Gualtieri, G. J.; Karnmlot, midity Environments. PNL-4284. Richland, WA: G. W.; Kelber, J.; Malm, D. L.; Sharpe, L. H.; Tier- Pacific Northwest Laboratories; 20 pp. ney, V. (September 1981). The Morphology and Corrosion Resistance of a Conductive Silver-Epoxy Daniel, J. L.; Coleman, J. E. [September 1980). “Pro- Paste.” Journal of Materials Science, Vol. 16 (No. gressive Changes in Microstructure and Composi- 9); pp. 2360-2368. tion During Degradation of Solar Mirrors.” Solar Reflective Materials; Proceedings of the Second Solar Reflective Materials Workshop; San Fran- Franz, H., inventor. PPG Industries, Inc., assignee. cisco, CA; 22-24 February 1980. Appears in Solar (16March 1976).“Selective Reflecting Metal/Metal Energy Materials, Vol. 3 (No. 1, 2); pp. 135-150. Oxide Coatings Using Surfactant to Promote Uni- form Oxidation.” U.S. Patent 3,944,440. Davies, J. M,, inventor. U. S. Secretary of the Army, assignee. (5 October 1971). “Flame Spraying Alu- Furukawa, K.; Kogyo, K. K.; inventors. Furukowa minum Oxide to Make Reflective Coatings.” U.S. Electric Co., Ltd., assignee. (29 May 1981).“Corro- 3,610,74 1. sion Inhibition of Copper, Silver, and Their Alloy.” Japan Kokai Tokoyo Koho 81, 62,973. Delameter, R. (October 1981). Private communica- tion to A. W. Czanderna. Gennari, Francois L. G., inventor. (15 December Della Mea, G,;Drigo, A. V.; Gothardi, V.; Nicoleteti, 1967).“Treatment of Gold, Copper or Silver Alloys F. (1978).“Near-Surface Analysis of Glasses Using with Silicones.” Fr. 1,505,506. Nuclear Techniques.” Silicates Industrials, Vol. 43 (NO.9); pp. 179-185. G bson, M. J.; Dobson, P, J. (October 1975), “Diffu- sion of Silver Through Epitaxial Films of Copper de Minjer, C, H.; van der Boom, P. F. J. (December and Nickel on Silver 111: The Formation of an 1973). “The Nucleation with SnCl,-PdCl, Solu- Equilibrium Layer of Silver.” Journal of Physics tions of Glass Before Electrolysis Plating.’’Journal F: Metal Physics. Vol. 5 (No. 10);pp. 1828-1835. of the Electrochemical Society, Vol. 120 [No. 12); pp. 1644-1650. G ostein, N. A.; Chavka, N. R. (1973).“Technological Dennis, W. E.; McGee, J. B. (September 1980). “Sil- Application of Auger Electron Spectroscopy.” icone Resins for Protection of First Surface Reflec- Journal of Testing and Evaluation, Vol. 1 (No. 3); tors.” Solar Reflective Materials; Proceedings of pp. 183-193. the Second Solar Reflective Materials Workshop; Sun Francisco, CA; 12-14 February 1980. Appears Godron, Y. (1976).“A Rational Bibliography on the in Solar Energy Materials, Vol. 3 (Nos. 1, 2); pp. Weathering of Glass Used in Building.” Verres 285-300. Refract. Vol. 30, No. 4, pp. 495-518, Doremus, R. H. (August 1968).“Exchange and Diffu- Goggin, R. (1981). Unpublished results of adhesion sion of Sodium and Silver Ions in Pyrex Glass.” measurements on aging Ag/glass mirrors. Golden, Physics and Chemistry of Glasses, Vol. 9 (No. 4); CO: Solar Energy Research Institute. pp. 128-132. Goggin, R.; Russel, P.; Pohlman, S. (1982). Corrosion Doremus, R, H.; Turkalo, A. M. (May 1976).“Electron in Solar Energy Systems.” Proceedings of the Elec- Microscopy and Optical Properties of Small Gold trochemical Society. Pennington, N. J. and Silver Particles in Glass.” Journal of Materials Science, Vol. 11 (No. 5);pp. 903-907. Gottardi, V.; Nicoleti, F.; Batteglin, G.; Della Meal G.; Ebert, J. (1982). “Activated Reactive Evaporation.” Mazzoldi, P. (1972).“On the Extraction Kinetics of Proceedings of the Society of Photo-Optical Instru- Sodium by Water in Different Float Glasses.’’ Riv. mentation Engineers: Volume 325-Optical Thin Stn. Sper. Vetro [Murano, Italy), Vol. 11; pp, 59-64, Films. Bellingham, WA: The Society of Photo- In Italian. Optical Instrumentation Engineers; pp. 29-38. Fan, J.C.C. (19 June 1981). “Sputtered Films for Gotz, Jiri. ( 1972). “Solution of Water in Commercial Float Glass.” Sb. Vys. Sk, Chem.-Technol. Praze Wave 1eng t h- Sele c t i v e Applications ” Th i n Sol i d . Techno]. Silikatu, Vol. LZ; pp. 255-261. In Films, Vol. 80 (Nos. 1,2, 3); pp. 125-136. Czec hoslavakian. Feldstein, N.; Weiner, J. A, [1972). “Contact-Angle Measurements of Tin-Sensitizing Solutions.” Jour- Gulf & Western Manufacturing Co., assignee. (9 Jan- nal of the Electrochemical Society, Vol. 119 (No. uary 1976). “Universal Faint Composition and 6);pp. 668-671. Objects Coated Therewith.” Brit. 1,532,691.

Physics and Chemistry of Silver/Glass Mirrors 39 Gupta, B. (January 1982). Major Event: Matrix Ap- 2,128,280. Also British Application 6486/1971; 10 proach for Testing Mirrors Provides Accelerated March 1971. and Abbreviated Method for Testing Durability of Jorgensen, G. (September 1980). Status Report on M i r r o r s .” SoI a r T h e r rn a I Tech n o J og y P r og r a m Monthly Status Report; 1-31 December 3983. Testing Advanced Mirrors. SERI/RR-225-1630. SERI/MR-251-1471, December. Golden, CO: Solar Golden, CO: Solar Energy Research Institute; 8 Energy Research Institute; v-vi. PP* Hamm, E. I., inventor. (20 December 1977). “Glass Karasek, J., inventor, (15August 1966).“Solution for and Mirror Making Process.” U.S. Patent 4,063,917. Silver Coating of Mirrors.” Czech, 119,610. Han, Y. H.; Kreidl, N. J.; Day, D. E. (January 1978). Khrusch, B. 1.; Agashkova, N. N.; Antonenko, A. L,; “Alkali Diffusion and Electrical Conductivity in Udovenko, V. F.; Lyulichev, A. N.; Dobrovol’s- Sodium Borate Glasses.” Journal of Non-Crystalling kaya, E. V. (1975). “Photostimulated Oxidation of Solids, Vol. 30 (No. 3); pp. 241-252. Silver Films.” Fiz. Khirn. Obrab Mater., Vol. 5; pp. 67-69. In Russian. Hartsough, L. D.; McLeod, P. S. (January/February 1977).“High-Rate Sputtering of Enhanced Alumi- Kienel, G. (6 March 1981).“Optical Layers Produced num Mirrors.” Journal of Vacuum Science and by Sputtering.” Thin Solid Films, Vol. 77 (No. 1, 2, Technology, Vol. 14 (No. 1);pp. 123-126. 3); pp. 213-224. Hass, G.; Heaney, J. B.; Herzig, H.; Osantowski, J. F.; Kobayashi, K. (April 1978). “Optical and EPR Stud- Triolo, J. J. (November 1975). “Reflectance and ies on Redox Interaction Layers of Ag’ and Cu+ Durability of Ag Mirrors Coated with Thin Layers Ions Diffusing into Soda-Lime Glass.” Physics and of A1203,Plus Reactively Deposited Silicon Oxide.” Chemistry of Glasses, Vol. 20 (No. 2); pp. 21-24. Applied Optics, Vol. 14 (No. 11);pp. 2639-2644. Kornppa, V. (1978).“The Effect of Heat Treatment on Hench, L. L.; Clark, D. E. (1982).“Surface Analysis of the Chemical Durability of Sodium Oxide - Sil- Glasses.” lndustrial Applications of Surface Anal- icon Dioxide and Sodium Oxide - Barium Oxide ysis. L.A. Casper and C. J. Powell, editors. Wash- - Silicon Dioxide Glass Surfaces.” G~QSSTechnol. VOl. 19, pp. 124-126. ington, D. C.: American Chemical Society;. __pp. 203-228. Kunin, N. F.; Shpilevskii, E. M. (1969). “Diffusion in Hepburn, W. D., inventor. (2 July 1980). “Manufac- Silver-Copper Thin Films.” Mater. Nauch.-Tekh. turing Reflector Having Silver or Gold Reflecting Konf. Fiz, Metal., Metalloved Obrab. Metal. Dav- Surface.” Britain U. K. Patent Application GB Ieniern., pp. 72-76. In Russian. 2,036,799. Kuznetsov, A. Y.; Maricheva, L. P.; Pospelov, B. A. Higginbotham, G. (1980). “The Weathering Proper- (October 1975).“Methods of Improving the Adhe- ties of Glass in Relation to Composition and Sur- sion of Metallic Silver Films to the Surface of face Treatment.” Proceedings of the 9th Austral- Glass and Quartz.” Soviet Journal of Optical ian Ceramic Conference; Sydney, Australia. Ken- Technology, Vol. 42 (No. 10);pp. 604-605. sington, New South Wales; pp. 168-173. Langley, R. C. (1974), “Absorbing and Reflecting Holland, L. (19561. Vacuum Deposition of Thin Films, Thin Films Deposited by Pyrolysis of Organome- New York: Wiley. tallic Compounds in Air.” Proceedings of the Sym- posium on Materials Science Aspects of Thin Film Holm, R.; Storp, S. (1976). “Surface Analysis of Ag- Systems and Solar Energy Conversion. PB-239- Sn Alloys by ESCA.” Journal Electron Spectros- 270, pp. 321-326. Available from NTIS. copy and Related Phenomena, Vol. 8 (No. 6); pp. 459-468. Larry, J. R., inventor; E. I. DuPont de Nemours and Isard, J. 0.;Patel, A. R. (December 1981). “ACompar- Company, assignee. (26 November 1975). “Metal- ison Between Weathering and Water Leaching lizations Comprising Nickel Oxide.” U. s. Patent Tests on Glasses of Simple Composition.” Glass 3,922,387. Technology, Vol. 22 (No. 6); pp. 247-250, Larry, J. R., inventor; E. I, DuPont de Nemours and James, J. (May 1982). SERI internal report. Golden, Company, assignee. [ 16 March 1976). “High- CO: Solar Energy Research Institute. Adhesion Conductors.” U.S. Patent 3,944,696. Jelli, A. (1974a). “Surface Degradation of Glasses.” Lebert, G. (January-February 1980). “Design and Silicates Industrials, Vol. 9 (No. 61; pp. 141-153. Manufacture of Reflecting Surfaces for Solar Pan- els.” Verres Refract., V. 34 (No. I);pp. 15-22. In Jelli, A. (1974b). “Surface Degradation of Glasses.’’ French. Silicates Industrials, Vol. 39 (No. 6);pp. 189-199. Jones, P. A,, inventor. (24 November 1972).“Forming Leidheiser, H. (1974). The Corrosion of Copper, Tin, Reflecting Surfaces on Articles.” French Dernande and Their Alloys, New York, Wiley.

40 Silver/Glass Mirrors for Solar Thermal Systems Liebig, J. (1835).Ann. D. Pharm., Vol. 14; p. 140. Morris, V. L. (September 1980). “Cleaning Agents and Techniques for Concentrating Solar Collec- Lind, M. A.; Buckwalter, C. Q.; Daniel, J. L.; Hartman, tors.” Solar Reflective Materials; Proceedings for J. S.; Thomas, M. T.; Pederson, L. R. (September the Second Solar Reflective Materials Workshop; 19793, Heliostat Mirror Survey and Analysis. San Francisco, CA; 12-14 February 1980. Appears PNL-3194. Richland, WA: Pacific Northwest Lab- in Solar Energy Materials, Vol. 3 (Nos. 1, 2); pp. oratories; various paging. 35-55. Lind, M. A., editor. (September 1980). Solar Reflec- tive Materials; Proceedings of the Second Solar Notoya, T.; Poling, G. W. (May 1979). Protection of Copper by Pretreatment with Bezotriazole.” Cor- Reflective Materials Workshop; San Francisco, rosion, Vol. 35 (No. 5);pp. 193-200. CA; 12-14 February 2980. Appears in Solar Energy Materials, Val. 3 (Nos. 1, 2); 346 pp. Nowotniak, K.; Wierzbicka, M.; Klayner, F,; Kas- Lind, M.A.; Chaudiere, D. A,; Dake, L. S.; Stewart, T. zuba, 2.; inventors. (20 December 19753. “Protec- L. (April 1982). Electroless Nickel and Ion-Plated tive Enamel for Mirrors.” Pol. 75.600. Protective Coatings for Silvered Glass Mirrors. Pearlstein, F. (June1979). “Selection and Application PNL-4257. Richland, WA: Pacific Northwest Lab- of Inorganic Finishes-VII-Metal Deposits for Spe- oratories; 36 pp. cialized Characteristics.” Plating and Surface Fin- Makijima, H.; Sakurada, Akira; Nagano, Kentaru; ishing, Vol. 66 (No. 6);pp. 42-50. Kuroyanagi, Takashi, inventors; Asaki Glass Co., Ltd., assignee. (18 November 1974). “Mirror Hav- Pederson, L. R.; Thomas M. T. (September 1980). ing Excellent Anti-Corrosive Property.” Japan “Characterization of New and Degraded Mirrors 74,42,897. with AES, ESCA, and SIMS4” Solar Reflective Materials; Proceedings of the Second Solar Reflec- Maruno, S. (1977).“Metallic Silver Discs Induced by tive Materials Workshop; San Francisco, CA; 12- Photosurface Deposition (PSD).”Journal of Non- 14 February 1980. Appears in Solar Energy Mate- crystalline Solids, Vol. 24 (No. 2); pp. 301-305. rials, Vol. 3 (Nos. 1, 21; pp. 151-167. Masterson, K. D.; Lind, M. A. (April 1983a)+Matrix Pederson, L. R. (19821. “Comparison of Stannous and Approach for Testing Mirrors-Part I. SERI/TP- Stannic Chloride as Sensitizing Agents in the Elec- 255-1504. Golden, CO: Solar Energy Research In- trolysis Deposition of Silver on Glass Using X-ray stitute; 35 pp. Available from NTIS. Photoelectron Spectroscopy.” Appears in Solar Energy Materials, Vol. 6 (No, 2); pp. 221-232. Masterson, IS. D.; Czanderna, A. W.; Blea, J.; Goggin, R.; Guiterrez, M.; Jorgenson, G.; and McFadden, J. Pellicori, S. F. (15 September 1980). “Scattering D. 0. (July 1983b). Matrix Approach for Testing Defects in Silver Mirror Coating.” Applied Optics, Mirrors-Part 2. SERI/TP-255-1627. Golden, CO: VOl. 19 (NO.18): pp. 3096-3098. Solar Energy Research Institute; 114 pp. Pettit, R. B.; Freese, J. M.; Arvizu, D. E. ( 1978).Specu- Matsumoto, K. (1981). “The Analysis of Carbon on lar Reflectance Loss of Solar Mirrors Due to Dust the Surface of Glass by ESCA.” Yogyo Kyokaishi. Accumulation.” Proceedings of the Seminar on Vol. 89 (No. 1);pp. 19-23. In Japanese. Testing Solar Energy Materials and Systems; Gai- thersburg, MD: 24 May 1978. Mt. Prospect, IL: Mattox, D. M. (January-February 1980). “Prepara- Institute of Environmental Science; pp. 164-168. tion of Thin Films for Solar Energy Utilization.” Journal of Vacuum Science and Technology. Vol. tt, C. W. (1981). “Thin Film Optical Waveguides 17 (NO.1); pp. 370-373. Fabricated by Ion-Assisted Techniques.” Thin Solid Films. Vol. 86 (Nos. 2,3);pp. 137-146. Mills, B. E. (September 1980). “Surface Studies of Corrosion in Second Surface Silver Heliostat Mir- tts, J. R.; Thomas, T. M.; Czanderna, A. W. (1984a) rors.” Solar Reflective Materials; Proceedings of Report in progress. Golden, CO: Solar Energy the Second Solar RefIective Materials Workshop; Research Institute (SERI/TR-255-2410). Sun Francisco, CA: 12-14 February 2980, Appears Pitts, Thomas, Czanderna, A$W, [ 1984b) in Solar Energy Materials, Vol. 3 (Nos. 1, 2); pp. J. R.; T. M.; 177- 184, Surface Analysis of Mirrors Made from Orga- nometallic Solutions. SERI/TP-255-1793, Golden, Minar, S,,inventor. (15 January 1975).“Protection of CO: Solar Energy Research Institute. Mirrors Against Degradation in Dusty and Wet Medium.” Czech. 156,592. Pohlrnan, S. L.; Russel, P. M, (September 1980),“Cor- rosion Resistance and Electrochemical Evaluation Morimoto, S.;Moori, Y. (1974).“Some Consideration of Silver Mirrors.” Solar Reflective Materials; of the Silvering Reaction.” Proceedings of the Proceedings of the Second Solar Reflective Mate- Tenth International Congress on Glass; Kyoto, rials Workshop; San Francisco, CA: 22-24 Febru- Japan; 8-12 July 1974. Tokyo, Japan: Ceramic ary 1980. Appears in Solar Energy Materials, Vol. Society of Japan; Part I, pp. 8-87 - 8-94. 3 (NOS.1, 2); pp. 203-213.

Physics and Chemistry of Silver/Glass Mirrors 41 Ramanathan, V. R. (1972). “Critical Conditions in Shelby, J. E.; Vitko, J., Jr-(May/June 1980a). “Surface Silvering on Glass.” Indian Chemical Journal, Vol. Characterization of Weathered Low-Iron Float 6 (NO+11); pp. 31-36. Glass.” Journal of Non-Crystalline Solids, Vol. 38 Rice, D. W.; Cappell, R. J.; Kinsolving, W.; Laskowski, and 39 (Part 11); pp. 631-636. J. J. (April 1980). “Indoor Corrosion of Metals.” Shelby, J. E,; Vitko, Jr., J.; Farrow, R. L. [September Journal of the Electrochemical Society, Vol. 127 1980b).“Characterization of Heliostat Corrosion.” (NO.4); pp.891-901. Solar Reflective Materials; Proceedings of the Rice, D. W.; Peterson, P.; Rigby, E. B,; Phipps, P. B. P.; Second Solar Reflective Materials Workshop; San Cappell, R. J.; Termourex, R. [February 1981). Francisco, CA; 12-14 February 2980. Appears in “Atmospheric Corrosion of Copper and Silver.” Solar Energy Materials, Vol. 3 (Nos. 1, 2); pp. Journal of the Electrochemical Society, Vol. 128 185-201. (NO.2); pp. 275-284. Shelby, J. E.; Nichols, M. C.; Smith, Jr., D. K.; Vitko, Roth, E. P.; Anaya, A. J. [November 1980). “The Jr.,J, (1981).“The Effect of Thermal History on the Effect of Natural Soiling and Cleaning on the Size Structure of Chemically and Vapor Deposited Distribution of Particles Deposited on Glass Mir- Silver Films on Glass.’’Materials Science Research, rors.” Journal of Solar Energy Engineering, Vol. VO~.14; pp. 579-589. 102 (NO.4); pp. 248-256. Sinclair, R. A., inventor: Minnesota Mining and Rymwid, Y.; Baker, I<. D., inventors; Hooker Chemi- Manufacturing Co., assignee. (17 July 1975).“Pro- cals and Plastics Corporation, assignee, (9 No- tection of Acid-Sensitive Metal Objects.” Ger- vember 1981).“Composition and Process for High many Offenlegungsschrift 2500018. Speed Electrodeposition of Silver.” U.K. Patent GB 2,086,940A. Sivertz, C.; Soltys, J. F., inventors. London Laborato- ries Ltd., Co., assignee. (1972).“Electrolysis Coat- Sakamoto, M., inventor. (26 July 1979).“Corrosion- ing with Silver.” Germany Offenlegungsschrift Resistant Mirrors.” japan Kokai Tokyo Koho 2162338. 79-94,516. Sanders, D. M.; Hench, L, L. (1973). “Environmental Steen, W. M. (1978).“Surface Coating Using a Laser.” Effects on Glass Corrosion Kinetics.” American Advances in Surface Coating Technology; London, England; 13-25 February 1978. Abington, England: Ceramic Society Bulletin, Vol. 59 (No. 9);pp. 662- 665, 666-669. Welding Institute; pp. 175-187. Saranov, E. I.; Kitaev, G. A.; Mokrushin, S. G. (1968). Stohr, J.; Jaeger, R. (JulyIAugust 1982). “Structural “Chemical Deposition of Mirror-Like Films of Studies of Schottky Barrier Formation by Means Nickel, Cobalt, and Copper on Glass.” Tr. Ural of Surface EXAFS: Pd and Ag on Si (111) 7 x 7.” Politekh. Inst. Vol. 170; pp. 108-112. In Russian. Journal of Vacuum Science and Technology, Val. 21 (NO.2); pp. 619-623. Schissel, P,; Czanderna, A. W. (September 1980). “Reactions at the SilveriPolymer Interface: A Re- Thomas, T. M.; Pitts, J. R.; Czanderna, A. W. (1983). view.” Solar Reflective Materials; Proceedings of Applied Surface Science, Vol. 15; p. 75. the Second Solar Reflective Materials Workshop; San Fruncisco, CA; 22-14 February 2980. Appears Thomas, T. M.: Pitts, J. R.; Masterson, K. D.; Czan- in Solar Energy Materials, Vol. 3 (Nos. 1, 2); pp. derna, A. W. (1984). Advanced Mirrars. SERI/ 225-245. RR-255-1629. Golden, CO: Solar Energy Research Institute. Schoen, J. M.; Poate, J. M.; Doherty, C. J.; Melliar- Smith, C. M. (November 1979).“Grain-Boundary Thomas, T. M.; Williamson, D.; Rainville, I). (198-1. Diffusion of Ag Through Cu Films.” Journal of “Role of Tin Complexes in +onding Silver to Applied Physics, Vol. 50 [No. 11);pp. 6910-6914. Glass: Mossbauer Studies.” Industrial Applica- tions of the Mossbauer Effect. New York: Plenum Schweig, 3. (1973). Mirrors: A Guide to the Manu- Press; in press. facture of Mirrors and Reflecting Surfaces. Lon- don, England: Pelham Books. Vaskelis, A.; Diemontaite, 0. (1975). “Electrolysis Silver Plating Solutions; 1. Silver Reduction in Shaisha, E. E.; Cooper, A. R, (1981).“Ion Exchange of Cyanide Solutions.” Liet, TSR Mokslu Akad. Darb., Soda-Lime Glass with Univalent Cations.” Journal Ser. B., (No. 4); pp. 23-30. In Russian. of the American Ceramic Society, Vol. 64 [No. 5); pp. 278-283. Vaskelis, A.; Demontaite, 0. (1976). “Electrochemi- cal Method for Adjusting Electrolysis SiIver Plat- Sharma, S. P. [December 1978).“Atmospheric Corro- ing Solutions.” Issled. Obl. Elektroosazhdeniya sion of Silver, Copper, and Nickel-Environmental Met. Mater. Resp. Konf. Elektrokhim. Lit. SSR, Test.” Journal of the Electrochemical Society, Vol. 14th.Vilnius,U,S.S.R.:Akad. NaukLit. SSR, Inst. 125 (NO.12); pp. 2005-2011. Khim. Khim. TekhnoI., pp. 192-196. In Russian.

42 SilverlGlass Mirrors for Solar Thermal Systems Vitko, Jr., J.; Shelby, J. E. (September 1980).“Solari- zation of Heliostat Glasses.” Solar Reflective Materials; Proceedings of the Second Solar Reflec- tive Materials Workshop; Son Francisco, CA; 12- 34 February 1980. Appears in Solar Energy Mate- rials, Vol. 3 (Nos. 1, 2); pp. 69-80. Vitko, Jr.,J.; Benner, R. E.; Shelby, J. E. (April 1982). SAND82-8627. Livermore, CA: Sandia National Laboratories. Viventi, R. V., inventor. General Electric Co., assignee. (1966). “Protection of Silver and Copper Sur- faces Against Corrosion.” Fr. 1,457,954. Watanabe, S.; Aritome, T.; Kuwabara, K.; Azeyanagi, T., inventors. Asaki Glass Co., Ltd., assignee. (8 August 1979). “Coating Materials for Mirrors.” Japan Kokai Tokyo Koho 79,100,414. Wiederhorn, S. M.; Lawn, B. R. (January/February 1979a). “Strength Degradation of Glass Impacted with Sharp Particles: I, Annealed Surfaces.” Jour- nal of the American Ceramic Society, Vol. 62 (Nos. 1, 2); pp. 66-70. Wiederhorn, S. M.; Lawn, B. R.; Hocky, B. J. (Novem- ber/December 1979b). “Effect of Particle Impact Angle on Strength Degradation of Glass.” Journal of the American Ceramic Society, Vol. 62 (Nos, 1, 2); pp. 639-640. Wille, B., inventor. Balzers Hochvakuurn G. M. b.II., assignee. (28January 1971).“Silver Mirror Protec- t ive Layers.” Germany Off enlegungs schrift 2028252. Williams, J. A.; Brungs, M. P.; McCartney, E. R. (April 1975). “Diffusion Mechanisms in Borosili- cate Glasses.” Physics and Chemistry of Glasses, VO~.17 (NO.2); pp. 53-56. Yoshihara, H.; Kiuchi, M., Aoki, T.; Umenura, S. (1977). “Some Properties of Ion-Plated Silver Films.” Nippon Kinzoka Gakkaishi. Vol. 41 (No. 10); pp. 999-1006. In Japanese. Zwissler, J. G.; Adams, M. A, (February 1981). JPL-PUB-81-16. Pasadena, CA: Jet Propulsion Laboratory.

Physics and Chemistry of Silver/Glass Mirrors 43 44 Silver/Glass Mirrors for Solar Thermal Systems Chapter 4 Mirror Product Testing Procedures

Introduction to Product Testing reliable data is to expose mirrors and mirror specimens to natural weathering for many years at the The optical performance requirements for mirrors intended solar installation sites. The obvious draw- used as solar concentrators has been detailed in ear- back is that sufficient data to meet the testing needs lier chapters of this document. Briefly, reflectors are not available for many years and probably not in used in solar concentrators should have a solar reflec- time to influence choices of components for a particu- tance of at least 90% and, preferably, as close to loo'% lar installation. In addition, real-time, long-term as possible. The mirrors must be highly specular so testing is insufficient for the near-termgoal of devel- that the reflected beam of light reaches the solar oping better mirrors. Long-term testing is most use- thermal receiver's absorbing surface without mean- ful in validating degradation models and life-predic- ingful loss due to scattering. In addition, in order to tive methodologies based on the short-term acceler- be cost-effective, these desirable optical properties ated or abbreviated tests, should be maintained over a 20-30 year service life during whicNhe component will be exposed to out- Abbreviated tests expose mirror materials to natural door weathering conditions that are diverse and weathering or controlled environments that simulate sometimes severe. natural weathering. The appropriate mirror proper- These requirements for solar concentrator applica- ties, such as reflectance and degradation products, tions are considerably more stringent than the per- are determined with a high degree of accuracy. The formance requirements that the established domes- results are extrapolated to longer time periods using tic mirroring industry has had to meet in the past. an appropriate model. The benefits of abbreviated The past performance requirements were directed testing are that data are collected over a short period toward mirrors used primarily for cosmetic and of time and under exposure conditions that should decorative purposes in an indoor environment. Vis- cause degradation by mechanisms identical to those ual appearance has been the primary concern and the found in deployed mirrors. The difficulties are that, basis for quality control and testing in the domestic in general, sensitive and accurate characterizations mirror industry. Such test criteria are of question- must be made and the degradation model must be able value for determining mirror performance and precise in order for the extrapolated lifetimes to be durability for the more demanding solar concentra- dependable. tor applications. In general, a formal product testing program should In accelerated testing, a mirror or mirror specimen is subjected to exposure conditions that are more severe have three fundamental objectives: (1)a near-term goal is to rank the performance of competing prod- than those seen in a natural weathering environment. ucts in order to meet advertising or procurement The benefit is that sufficient data to generate a substantial amount of degradation can be collected in a requirements; (2)an intermediate goal is to develop a testing methodology for predicting the useful service short period of time. This allows a more precise life of solar concentrator mirrors used in various mathematical modeling of the degradation rates. The environmental regions in order that more reliable drawback of accelerated testing is that under severe analyses of their cost-effectiveness and system eco- exposure conditions, a different set of degradation nomics can be performed: and (3)the third goal is to mechanisms from those causing real-life failure may obtain a basic understanding of mirror degradation be invoked; although the mathematical model may be mechanisms so that the mechanisms can be passi- well defined, it may be highly inaccurate in determin- vated or designed out of the mirrors in order to obtain ing service life for solar concentrator mirrors. Never- outstanding durability. A basic understanding of the theless, accelerated testing, because of the quick degradation mechanisms provides the knowledge acquisition of data, is the preferred method of indus- upon which more appropriate testing procedures can try and research institutions for testing and develop- be developed. Also, limits on the reliability of such ing mirrors and test methodologies. Only after the procedures can be set. results from accelerated testing are found to correlate well with real-time, long-term, natural weather- A thorough and complete testing program that would ing, can one say that a reliable test methodology fulfill the needs embodied in these goals would exists and that the degradation mechanisms are well proceed on several levels. The level giving the most understood.

Mirror Product Testing Procedures 45 Existing Test Methodologies mine their suitability for such applications. The environments used lo degrade the performance were The principal market for the domestic mirror indus- high fluxes of extreme ultraviolet light and energetic try is for cosmetic and decorative mirrors for which H proton and electron beams. For terrestrial applica- successful product need only have good visual ap- t ions, such causative agents are inappropriate. The pearance and a 10-20 year lifetime in rather benign data Iron1 the ultraviolet exposure of silvered fused environments. High humidity and condensation does, silica are pertinent to testing of silveriglass mirrors however, occur often in bathrooms and has resulted and will be discussed later. During the mid-l960s, and in forming black spots and edge blackening. The in the space program, several ASTM standards for industry standard for quality control testing is the optical characterization of materials were developed. salt spray tesl. In this test, mirrors are subjected to a fog of salt water “1.S. Government test DDM411 or Testing Requirements for Solar ASTM B117).?’he fog condenses on the specimen and Concentrators drips off. Visual inspection 01 the degraded mirrors by trained Because of the unique requirements for solar concen- individuals is used to measure the mirror perform- trator mirrors, testing programs that have been ance qualitatively. Although this may serve for qual- established for other industry or consumer uses have ity control and mirror comparisons, it is a poor sub- been found inadequate. A more thorough and scien- stitute for quantitative measures of performance tific approach to testing is necessary before lifetime needed for solar concentrator engineering. The na- predictive methodologies and advanced, more duration’s largest organized consumer, the federal govern- ble mirrors can be produced. Appropriate acceler- ment, does apply specifications and standards for ated tests are required in order that data be obtained procurement activities. However, these standards in a reasonable length of time; long-term natural (GGG-M-35Oa, DD-M-O0411b, and MIL-STD-105) weathering tests are necessary to validate degrada- refer only to attributes such as appearance and con- tion models in predictive methodologies. struction and to the statistical procedures to be used A first step in an accelerated testing program is in acceptance or rejection of products. based on the best knowledge of the causative agents Another consumer of mirrors subjected to outdoor for degradation. The interactions that lead to mirror weathering characteristics is the automobile indus- degradation and corrosion are probably of an elec- try, whose primary concern is durability and ap- trochemical nature with physical diffusion through pearance. High reflectance is not required. In fact, it the thin films and interfaces of the mirror’s structure. has been found that somewhat lower reflectance, The causative agents may be internal in nature; that which provides contrast for the mirror in the ambient is, they may be produced by contaminants entrained background, is desirable. Hence, modern automobile during the mirroring process, electrochemical inter- mirrors are metallized with chromium or nickel action with the paint-protective coatings, or from the instead of silver. Apparently, the industry also uses basic electrochemical interaction of the various metal a salt spray test to monitor product quality. layers and glass surface. The causative agents may also be from external sources such as liquid or vapor The German autoinobile industry has reportedly forms of water, environmental pollutants, elevated used both a salt spray (DIN 50021) and a test that temperature, ultraviolet light, or temperature cycling. uses alternating damp heat and sulfur dioxide vapors Degradation may be, in fact, a combination of exler- (DIN 50018) for chromium mirrors. The criteria for nal and internal causative agents with external characterizing mirror degradation are not identified. agents such as high temperature and water vapor It is interesting to note that before World War 11, a acce 1 era t i ng Ihe i n tern a 1 me c h a n i s in s . substantial number of silvered glass mirrors were fabricated for use in the auto industry. Testing procedures and qualifications used at that time have Characterization of Mirror Performance not been determined by the authors. Test equipment and Degradation and sophisticated analysis techniques such as those used in modern surface analysis and ultra-high The fundamental performance parameter for solar vacuum technology were not available. The degrada- mirrors is the specular solar reflectance. If the mirror tion problems, if they existed, were designed out of maintains good specular reflectance, changes in other the system by switching to chromium metallization. optical, physical, or mechanical properties are less important. In a formal testing program, however, any One application for solar reflectors that has received measurement that leads to elucidation of the clegra- considerable testing is the thermal control of space dation mechanisms, which provides an extremely vehicles. In this application, high solar reflectance sensitive measure of the onset of degradation, or was useful to reduce solar heating of spacecraft. In which can be related easily to specular solar reflec- addition, high emittance was desirable to dissipate tance, is worthy of consideration. energy, Extensive testing (Marshall 1968; Hass 1970; Hollingswort h 1977; and Cunningham 1970) was Preferred characterization techniques are sensitive, performed on candidate materials in order to deter- nondestructive, and relatively inexpensive to per-

46 SilverlGlass Mirrors for Solar Thermal Systems form. Optical reflectance measurements generally have been scrutinzed in several studies. Several elec- meet these requirements. Furthermore, they are di- tron and ion microscopy techniques, together with rectly related to the fundamental parameter of the the rationale for their use and some results, are sum- solar reflectance. Several forms of reflectance mea- marized in Table 4-2. surements have been used to characterize mirror degradation. Diffuse reflectance measurements using Photomicroscopy is a valuable technique for ob- integrating sphere reflectometers are sensitive in serving mirror degradation. Its primary use is in detecting early degradation. But because of low sig- viewing the degradation and enabling one to identify nal levels, they appear to offer little advantage over morphological changes in the various components of specular reflectance measurements. However, mea- the mirror structure. Microscopy is also valuable for surements of total integrated scatter using a Coblentz observing the light-scattering properties of degraded sphere could be performed simultaneously with the mirrors. However, the use of these techniques as specular reflectance measurement and would seem quantitative measures of mirror degradation is lim- to offer useful additional data on the characteristics ited because of difficulty in maintaining appropriate of degraded mirrors. Table 4-1 summarizes the ap- exposure and detection levels. The techniques are plicable reflectance techniques, their rationale, and also too time consuming to be used extensively for the results of their application to the mirror degrada- quality assurance testing. tion problem. Optical photomicroscopy has been used qualitatively In contrast to the macroscopic measurements repre- to characterize mirror degradation (Lind 1982a; Dake sented by the optical reflectance techniques, micro- 1982;Masterson 1983a). The necessity to observe the scopic variations in the degraded mirror specimens silver/glass interface through the glass superstrate

Table 4-1 Reflectance Techniques for €valuating Loss in Mirror Performance

Techniques Rationale Limitations/Results

Hemispherical reflectance Search for surface plasmon effects in Time consuming for spectral scan roughened silver films for 300- to Revealed short wavelengths most 800-nm spectral scans sensi tive . Hemispherical solar Hemispherical reflectance is Fast, but not as accurate as 1 and 5. reflectance using portable spectrally averaged ref Iec tometer Surface enhanced Raman Monitor small chemical changes in Difficult to use; signals associated spectroscopy (SER) the si Iver/g lass interface with agglomerated films were rep0 rted. Diffuse reflectance Obtain spectral information sensitive Not as sensitive as hemispherical to scattering centers reflectance when using typical i ntegrat ing spheres, 5. Spectral reflectance Search for Bruggeman and Garnett Time consuming; same as 1; allows roughening effects using 300-2500 calculation of solar-weighted nm spectral scans ref Iectan ce. 6. Specular reflectance Direct measure of the performance Difficult to use consistently. Results using portable characteristic of greatest interest corn parable to other reflectance ref Iecto mete r techniques. 7. Differential laser scans Monitor differences in reflectances Time consuming. Inconsistent at two laser frequencies to look for correlation with 350-nm vs 650-nm chemical or structural changes sources. 1 :lo4sensitivity. 8. Darkf ield photography Measure of light-scattering centers Fast, permanent record. Definite (1OOX) trends that correlate to hemispherical reflectance. 9. Photography (UV) Monitor local regions of surface Hard to do with conventional light plasmon enhancement sources: no enhanced regions.

Mirror Product Testing Procedures 47 Table 4-2 Results Using Surface Analysis Techniques for Evaluating Mirror Degradation

Technique Results

Auger Electron Spectroscopy (AES) Detected significant diffusion of Cu, CI, and oxygen through the silver film to the glass interface. Scanning Electron Microscopy (SEM) Revealed agglomeration of silver in corroded areas of mirrors. Ion Scattering Spectroscopy (ISS) Elucidated the relative amounts of Ag and Sn bonded onto glass after sensitizing and silvering without reducers. Secondary Ion Mass Spectroscopy (SIMS) Provided first data that suggested Ag-Si bonds are adhesion for glass. X-ray Photoelectron Spectroscopy (XPS) Identified nature and density of silver/glass bonding sites in uncorroded mirrors and their dependency on preparation techniques.

limits the magnification to 1OOx-2OOx. Nevertheless, Figure 4-1 shows typical nuclei that formed during the formation and growth of pinholes or other solid accelerated testing. Attempts to quantify the early phases in the silver layer is readily observable and stages of degradation by counting the density of provides qualitative information on the nature and light-scattering defects, or by determining the frac- degree of degradation in the thin silver film. Both tional area from which the light-scattering signal dark field and lightfield illumination have been used. arises, suffered from the difficulty in reproducibly

Figure 4-1 Darkfield Photomicrographs (1OOx) of Degraded Commercial Mirror [The exposure periods are as follows: (top, left) unexposed; (top right) after two weeks; (bottom left) after four weeks; (bottom right) after eight weeks.] (Masterson 1983b)

48 Silver/Glass Mirrors for Solar Thermal Systems setting the gray scale. This is a threshold level for concentrator test facilities (ASTM E838-8) and ex- detection that is critically sensitive to instrument- posed to solar radiation levels of eight times the nat- operating temperatures and specimen illumination ural exposure level. During such exposure, speci- levels for commercially available microdensitome- mens accumulated total radiation amounts equal to 25 ters and image analysis systems (Masterson 198313). years of natural weathering. Modeling of the initial results for degradation in reflectance by Honeywell Similar difficulties are experienced in the use of pho- (Rausch 1980), using a modified Weibull function, tographic film to quantify the nature and density of was successful in describing the onset of degradation defects in the film as a function of exposure time to in reflectance. the degrading environment. The principal cause for this difficulty is the variation in film and exposure Long-term exposure and carefully standardized op- characteristics, which is apparently caused by poor tical measurements are necessary before this prelim- control of shipping and storage conditions and times inary analysis can be verified adequately. Expo- for the films. One further drawback of any attmept to sure tests for shorter periods of time have been quantify mirror degradation by photomicroscopy conducted by McDonald-Douglas [Morris 1980) on techniques is that the degradation in commercially test racks located at a number of sites selected for produced silveriglass mirrors varies substantially solar industrial process heat demonstration projects. from one location on a specimen to another, and Although degradation in the silver reflector layer within a few millimeters. Thus, great statistical was observed and measured, more significance in uncertainties appear unavoidable. this study was placed on soiling and cleaning problems that were site specific. A common problem in The presence and quantification of light-scattering many of the mirror degradation studies is a lack of defects in mirror specimens can be more easily ac- standardization in the reflectance measurements, complished using sensitive diffuse reflectance mea- and thus, it is difficult to compare the results of the surements such as the total integrated scatter method different experiments. described by Bennett (1978).The application of this method to quantifying degradation of mirrors for Another form of natural weathering tests, which is a solar concentrators under accelerated exposure or good source of information, is the deployment of full- natural weathering conditions would appear promis- scale heliostat modules at a number of test sites and ing, but, as yet, it has not been attempted. demonstration projects. Degradation observed in these modules provided the impetus for the laboratory degradation studies. It has been useful in estab- Results of Exposure Testing lishing appropriate laboratory exposure and test procedures (Burolla 1980; Daniel 1980; and Shelby Various natural weathering and accelerated expo- 1980). Degradation in these modules was severe in sure tests have been performed in attempts to rank cases where liquid water was trapped in the struc- the durability of candidate silver mirrors and to ture. This observation and subsequent tests have led assess their long-term performance in solar applica- to a significantly better understanding of the role of tions. Accelerated testing has been carried out in water in the degradation process. Another important both liquid- and vapor-phase environments. contribution of outdoor weathering programs is that specimens with long-term outdoor exposure are be- coming available for detailed analysis and for com- Natural Weathering parison of degradation mechanisms to those observed in laboratory-accelerated tests. Natural weathering tests are necessary to identify the degradation processes that should be accelerated during laboratory level durability studies. They are Liquid Phase Testing also necessary in order to provide long-term weat h- ering data that can be correlated with short-term Liquid phase testing of solar mirror materials has results in order to validate short-term testing and been conducted in several configurations. Specimens degradation models. Outdoor weathering tests were have been completely immersed in synthetic sea- begun in the solar energy program by Honeywell water (Pohlman 1980) and in deionized water; both, (Rausch 1978) and by the Illinois Institute of Tech- with and without dissolved oxygen (Vitko 1982).The nology Research Institute (Gilligan 1979 and 1980) at latter tests were conducted at temperatures up to various sites within the United States. Many of the 90°C and produced pinhole formations similar to materials were exposed at DSET Laboratories, Inc. that observed in field-exposed mirrors except for an near Phoenix, Arizona. DSET Laboratories is contin- absence of agglomerated silver particles in the cor- uing the exposure and evaluation of several of the roded spots (Figure 3-6 in Chapter 3). Degradation reflector materials included in the earlier studies. was substantial only when oxygen was dissolved in the water. This result is consistent with an earlier In exposure tests, specimens of mirror materials are experiment where a cup, partially filled with water, mounted on exposure racks and their reflectance is was bonded to the back side of a mirror specimen measured periodically. In the Honeywell experiments, which was then placed in a vertical frame (Burolla specimens were also mounted at the focus of solar 1980).The results of this partial immersion indicated

Mirror Product Testing Procedures 49 corrosion. Degradation was more severe when the Vapor Phase Test ratio of the volume of air to volume of water was large than when the ratio was small. For large ratios, Concern that liquid phase testing of solar mirror the amount of available oxygen is substantially specimens might be too severe and that it does not higher. realistically model the actual outdoor exposure con- AIternate imrnersian tests and salt spray exposures ditions has led several researchers to conduct an using techniques similar to that of many mirror accelerated exposure test where the mirror speci- manufacturers have also been used in laboratory mens are subjected to environments characterized by studies of mirror degradation. These tests were done water and pollutants maintained in a vapor phase. to help evaluate liquid cell electrochemical tech- Strong vapors of sulfuric acid and hydrochloric acid niques for ranking the durability of commercially degraded mirrors quickly (Coyle 1982).Hydrochlo- made silver mirrors (Pohlman 19801. The results of ric acid vapors were more effective in causing rapid the electrochemicaI testing were encouraging since a degradation. The acid vapors attacked the backing good correlation in durability was obtained with paint protective layer on the mirror specimen with rankings made using other tests. Furthermore, the some diffusion of ions to the metalizing layers. electrochemical testing can be done in a few hours as Whether or not such a severe environment is appro- compared to many days required for other acceler- priate for ranking mirror performance or predicting ated lifetime testing methods or to many years mirror lifetime is yet to be established. A similar test required for real-time weathering studies. (DIN 500181, in which specimens are alternately subjected to high concentrations of sulphur dioxide and In a reported test (Pohlman 19801, a scratch was 100% humidity at 4OoC, is reported to be used by made through the paint and metal layers so that the some German auto makers for testing rearview mir- interfaces were exposed to the electrolyte in the cell; rors. No further data on its use by the mirror industry then the electrochemical potentials and corrosion or any test results were available. currents associated with the metalizing layers were characterized. Since a field study of mirror degrada- In an effort to study the effects of the solar environ- tion must also include the ability of the backing paint ment on mirror degradation more systematically, an to protect the metalization from corrosion, additional extensive study associated with individual and mul- studies to determine the permeability of the paint to tiple degrading agents was initiated under SERI’s water vapor and ionic transport would be necessary leadership (Masterson 1983a).In planning and coor- to supplement the electrochemical test before a real- dinating this project, SERI was assisted by person- istic lifetime predictive methodology could be estab- nel from Sandia National Laboratories (SNL) and lished. Nevgrtheless, the approach looks promising Pacific Northwest Laboratories (PNL].A considera- and would warrant further consideration in future ble amount of laboratory experimental work and studies of the mirror degradation process. mirror characterization was conducted (Masterson 1983a; 1983b; Lind 1982a and 1982b; Dake 1982). In these studies, mirrors produced by several commer- Salt Spray Test cial mirroring companies were subjected to environments containing high levels of solar or ultraviolet Standardized salt spray tests (ASTM B117-73) are radiation, SO, and NO, pollutant gases, relative used by the mirror industry (Bomar 1981) to assess humidity, mechanical stress, and elevated tempera- the durability of developmental mirror systems and tures. These agents acted individually or in multiples for quality control of production mirrors. Unfortu- of two, three, and four to test for synergistic accelera- nately, many of the results of these studies and the tion of degradation. The results are detailed else- experience gained from such tests are considered where (Masterson 1983a, 1983b; Dake 1982; Lind proprietary by the various manufacturers and none 1982a and 1982b] and will only be summarized here. of the work was found referenced in the literature survey. Qualitative information on degradation rates The most important results of these studies are that obtained from industry representatives, the results mechanical stress at levels close to those necessary of the liquid- and vapor-phase exposure testing, and for glass fracture, ultraviolet light at levels greater the degradation models proposed elsewhere in this than 10 times those in the solar spectral irradiance report indicate that salt spray exposures are worthy below 310 nm, and pollutants at levels near 100 times of further study. This test has the necessary elements the maximum allowed by EPA for one-hour human of high humidity, liquid phase water with substan- exposure caused little or no degradation in mirror tial dissolved oxygen, and electrolytes. It has the specimens after exposure of up to eight weeks. There potential to be used at elevated temperatures which were also no synergistic effects observed where the makes it an attractive method for accurate acceler- exposure to these agents caused any degradation ated testing of solar mirrors. Any future program to that was more rapid than the exposure to the high assess mirror degradation and predict mirror service temperature and humidities alone. life should give the salt spray test serious consideration for either direct application in the testing pro- In carefully controlled experiments (Masterson1983a, cess or should modify it to more nearly meet the 1983b; Lind 1982a, 1982b), eight weeks exposure to requirements of testing for solar mirror materials. 80°C and relative humidity greater than 75% resulted

50 Silver/Glass Mirrors for Solar Thermal Systems in a degradation in hemispherical solar reflectance of As part of this study, the ability of several optical up to 0.15 reflectance units in typical silver/glass characterization methods to detect early stages of mirrors [Figure 4-2). Degradation was accelerated at degradation and to resolve differences statistically in the high relative humidities where monolayers or degradation rates was also assessed. Many of the op- greater of condensible water formed on the surfaces tical reflectance measurements used were sensitive in of specimens. It is suspected that corrosion rates detecting mirror degradation. Stat istical variations were not humidity-dependent at lower relative hu- observed in most of the data appeared to arise from midities, but the verification of this hypothesis real differences in mirror degradation rates rather would require more data at the levels of relative than from uncertainties in the optical measurements. humidities of interest. Mirror degradation rates appear to vary substan- One unexpected observation was that specimens tially across distances of only a few millimeters on exposed to the pollutant gases showed less degrada- the same mirror surface. For silver/glass mirrors, the tion than when exposed to water vapor alone. Among spectral reflectance shows considerably more rapid several possible explanations for this, the most degradation at wavelengths below 500 nrn (Figure likely is that the nitrogen gas purged oxygen from the 4-3). A wavelength of approximately 400 nrn would exposure environment and thus prevented any appear to be optimal for monitoring the degradation oxidation-reduction reactions, which may be respon- of both hemispherical and specular reflectance in sible for much of the observed mirror degradation. silver/glass mirrors. A single measurement of specular reflectance at that wavelength combines degradation effects due to the scattering of light by roughening of the silver metal layer and increased absorptance caused by corrosion products and pinholes, I Legend T

High Temperature Exposures

One characteristic observed by photo and scanning microscopy of field-degraded mirror modules was 0.07 - the apparent agglomeration of silver into rnicro- scopic size particles in the center of the severely 0.06- corroded areas [Shelby 1980).Since silver is known 0.05 - to agglomerate readily at temperatures above 100°C- 150OC (Czanderna 1975; Presland 1972; Rhead 14633, 0.04- several laboratories attempted to utilize high-temperature exposure as a mechanism for studying the durability of field-exposed mirrors and to develop appropriate lifetime testing methods (Shelby 1980; Dake 1982; Masterson 1983b). In one study, exposure of a bare silver film in the -o,o-it & Temperature, O C ambient laboratory atmosphere at 150°C produced substantial agglomeration in a few hours. This is consistent with studies of silver agglomeration and Figure 4-2 Dependence of Degradation in Reflectance on Tem- was used to hypothesize that silver diffusion plays perature far High and Low Relative Humidity Using an important role in the degradation process. In the Select, Quality Commercial Mirrors (Masterson 1983b)

Figure 4-3 Spectral Hemispherical Reflectance for Different Exposures of a Commercial Mirror [Values for the exposure periods and the solar-weighted reflectance (p,) are shown in the legend.] (Masterson and Lind 1983a)

Mirror Product Testing Procedures 51 SEKI studies (Masterson 1983b3, a commercially edge corrosion and edge delamination in some spec- produced wet chemical-processed, silvered mirror imens. However, this does not degrade the specular was exposed to temperatures between 4OoC and reflectance of the intact central portions of the spec- 200°C by inserting one end of the specimen into a imens as quickly as the steady temperature condi- laboratory furnace while the other end was cooled. tion. It appears that the length of time at high Degradation was substantial at the high-temperature temperature is a critical factor in degradation. end of the specimen. Decreases in hemispherical Another finding from these controlled exposures is reflectance at 400-nm wavelengths were measured along the mirror specimen and correlated to mirror that specimens could be removed from the test chamber, characterized, and replaced without affect- temperature, ing the observed degradation rates. Also, simultane- The data were used to produce an Arrhenius plot ous exposure to multiple environments indicated from which an activation energy for agglomeration that the presence of ultraviolet light or typical gase- in the silver film of approximately 30 kJ/mole was ous pollutants does not significantly accelerate the determined. This activation energy was only about observed degradation. There were no synergistic one half of that previously measured, 69 kJ/mole effects observed between the various exposure para- (Presland 1972), for a bare film exposed to air. The meters; that is, the combined degradation rates were reason for the extremely low activation energy is not just the sums of the degradation rates observed for completely understood, but it may be caused by the the individual parameters. prcsence of impurities in the silver layer accumu- In these studies of mirror degradation, great varia- lated during the manufacturing process or by the tions in degradation rates were observed between thermal gradient along the 30-cm long specimen. specimens of a single mirror type and even between Since agglomeration is severe and takes place macro- different areas of a single specimen. This appears to scopically at temperatures exceeding 9O0C, it is con- be caused by non-uniformity of thickness and com- cluded that all future accelerated weathering tests to position of the various reflecting and protective lay- determine mirror durability and mirror ranking ers that constitute a complete mirror. The non- should be performed at a temperature less than 80°C uniformity is probably a consequence of the wet until studies can be made to demonstrate that no chemistry mirroring process as used by all commer- fundamental mechanistic changes occur above 80°C. cial silver/glass mirroring companies. The conse- A c c o r d i ng I y , s y s t ema t i c a n d c a re fully c o n t r o 1led quence of this to accurate durability testing is that high- t e m p e r;i t u re expos u re s and subseq u e n t deter - many specimens and measurements are needed to mination of the activation energies associated with obtain results with high statistical precision. the accompanying degradation may yield valuable insight and quantitative data on the degradation of No adequate testing scheme has been evolved for predicting mirror lifetime and durability. Accurate c o m me r c i a 11 y produced mirrors , reproduction of all degradation phenomena observed in field-exposed mirrors has not been accomplished Summa ry during accelerated laboratory testing. Nevertheless, substantial progress has been made in understand- Numerous testing procedures have been invoked ing many of the causes of degradation and in design- over the past few years to assess the durability of ing appropriate accelerated tests for determining silver/glass mirrors and to determine models for mirror durability. Substantial work has yet to be their degradation in the various exposure environ- done to obtain results for long-term, outdoor, accel- ments. The results, to date, indicate that the most erated exposures; to determine the rate equations for important parameters to accelerate mirror degrada- the degradation process; and to validate degradation tion are elevated temperature, humidity (greater models. Analytical techniques for characterizing the than 80% and high enough that condensible water is optical properties of degraded mirrors are now better present on the back surface of the mirror), and an defined, Surface analysis techniques to identify mir- abundance of oxygen to participate ih the oxidation- ror corrosion products and interface interactions reduct ion react ions. have been a useful tool in arriving at the present understanding of the causes of mirror degradation. Spectral reflectance degrades most rapidly at the short wavelengths; near 400 nrn [see Figure 4-3) for References silver/glass. Bennett, H. E. (1978). “Scattering Characteristics of In exposure to high humidity and high temperature, Optical Materials.” Optical Engineering, Vol. 17 etching of the front surface of the glass superstrate is (NO.5); pp. 480-488. severe for some types of glass. When the purpose of the exposure is to test the stability of the optical prop- Bomar, S.; Keith, D.; Maddox, I<.; Walton, J. D. (1981). erties of the metal reflecting layer, the glass surface Workshop Proceedings: Durability of Reflecting must be protected from the corrosive environment. Surfaces Used in Solar Heliostats, Georgia Insti- tute of Technology, Engineering Experiment Sta- Temperature cycling at high humidity causes severe tion, Atlanta, GA, 1981.

52 Silver/Glass Mirrors for Solar Thermal Systems BuroIla, V. P. (September 1980).“Deterioration of the Efficiency and Solar Applications. Edited by Carl Silver/Glass Interface in Second Surface Solar M. Lambert. Los Angeles, CA: 28-29 January 1982. Mirrors,” Solar Reflective Materials; Proceedings Bellingham, WA: SPIE; pp. 78-83. of the Second Solar Reflective Materials Work- Lind, M. A,;Chaudiere, D. A.; Dake, L. S.; Stewart, T. shop; San Francisco, CA; 12-14 February 1980. L. (April 1982bj. Electrolysis Coatings for Sil- Appears in Solar Energy Materials, Vol. 3 (Nos. 1, vered Glass Mirrors. PNL-4257. Richland, WA: 2); pp. 117-126. Pacific Northwest Laboratories; 36 pp. Coyle, R. T.; Barrett, J. M.; Call, P. J. (March/April 1982). “Durability of Silver-Glass Mirrors in Marshall, K. N.; Breuch, R. A. (September 1968). Moist Acid Vapors.” Appears in Solar Energy “Optical Solar Reflector: A Highly Stable, Low oc,/e Spacecraft Thermal Control Surface.” Jour- Materials, Vol. 6 (No. 3); pp. 351-373. nal of Spacecraft, Vol. 5 (No. 9); p. 1051. Cunningham, F. G.; Bean, B. L.; Park, S. G. (19701. Masterson, K. D*; Lind, M. (April 1983a). Matrix “Ultraviolet and Charged Particle Degradation of Approach for Testing Mirrors - Part 1. SERI/TR- Aluminum and Silver Coated FEP Teflon Second 255-1504. Golden, CO: Solar Energy Research In- Surface Mirrors.” Space Simulation. NBS Special stitute; 35 pp, Publication Number 336. Washington, D,C.: Na- tional Bureau of Standards; pp. 345-358. Masterson, K.; Czanderna, A. W.; Blea, J.; Goggin, R.; Jorgensen, G.; Guiterrez, M.; and McFadden, J. D. Czanderna, A. W., editor. (1975). Methods of Surface 0. (July1983b). Matrix Approach for Testing Mir- Analysis. New York: Elsevier Science Publishing rors - Part 2. SERUTR-255-1627. Golden, CO: Company; 481 pp. Solar Energy Research Institute. Dake, L. S.; Lind, M. A. (April 19821. The Effect of Morris, V, L. (September 1980). “Cleaning Agents Exposing Two Commercial Manufacturers’ Second and Techniques for Concentrating Solar Collec- Surface Silver/Glass Mirrors to Elevated Temper- tors.” Solar Reflective Materials; Proceedings for ature, Mechanical Loading, and High Humidity the Second Solar Reflective Materials Workshop; Environments. PNL-4282, Richland, WA: Pacific San Francisco, CA; 12-14 February 1980. Appears Northwest Laboratory; 20 pp. in Solar Energy Materials, Vol. 3 (Nos. 1,Z); pp. Daniel, J. L.; Coleman, J+E. [September 1980). “Pro- 35-55. gressive Changes in Microstructure and Composi- Pohlman, S. L.; Russell, P. M. (September 1980).Cor- tion During Degradation of Solar Mirrors.” Solar rosion Resistance and Electrochemical Evaluation Reflective Materials; Proceedings of the Second of Silver Mirrors.” Solar Reflective Materials; Solar Reflective Materials Workshop; San Fran- Proceedings of the Second Solar Reflective Mate- cisco, CA; 22-14 February 1980. Appears in Solar rials Workshop; San Francisco, CA; 12-14 Febru- Energy -Materials, Vol. 3 [Nos. 1, 2); pp. 135-150. ary 1980. Appears in Solar Energy Materials, Vol, 3 (NOS.1, 2); pp. 203-214. Gilligan, J* E. (1979).“Exposure Testing and Evalua- tion of Solar Collector Materials.” Proceedings of the Annual Solar Heating and Cooling Research Presland, A. E. B.; Price, G. L.; Trimm, D. L. (1972). and Development Branch Contractors’ Meeting; “Kinetics of Hillock and Island Formation During Washington, D. C.; 24-27 September 1978. Wash- Annealing of Thin Silver Films.” Progress in Sur- ington, D. C.: Department of Energy; pp. 78-79. face Science, Vol. 3 (Part I); pp. 63-96. Gilligan, J. E. (April 1980). Exposure Testing and Rausch, R. A,; Gupta, B. P. (1978). “Exposure Test Evaluation of Solar Collector Materials. DOE/CH- Results for Reflective Materials.” Proceedings of 90034-TI. Chicago, IL: IIT Research Institute, 92 the Seminar on Testing Solar Energy Materials PP. Systems, Mt. Prospect, IfL: Institute of Environ- mental Science; pp. 184-187. Hass, G.; Hunter, W, R. (September 1970). “Labora- tory Experiments to Study Surface Contamination Rausch, R. A. (October 1980).A Correlation Study of and Degradation of Optical Coatings and Mate- Accelerated and Nonaccelerated Exposure Test rials in Simulated Space Environments,” Applied Results for Three Solar Reflective Materials. Optics, Val. 9 (No. 9);pp. 2101-2110. SERUR-8270-1-D, Golden, CO: Solar Energy Re- search Institute; 38 pp. Hollingsworth Smith P.; Gureo, H. (1977). “Silicon Dioxide as a High Temperature Stabilizer for Silver Rhead, G. E. (1963). Acta Metallurgica; Vol. 11; p. Films.” Thin Solid Films, Vol. 45; pp. 159-168. 1035. Lind, M. A.; Chaudiere, D. A,; Stewart, T. L. (1982a). Shelby, J. E.; Vitko, Jr., J.; Farrow, R. L. (September “Electrolysis Nickel and Ion-Plated Protective 1980). “Characterization of Heliostat Corrosion.” Coatings for Silvered Glass Mirrors.” Proceedings Solar Reflective Materials; Proceedings of the of the Society of Photo-Optical Instrumentation Second Solar Reflective Materials Workshop; San Engineers: Volume 324-Optical Coatings for Energy Francisco, CA; 12-24 February 1980. Appears in

Mirror Product Testing Procedures 53 Solar Energy Materials, Vol. 3 (Nos. 1, 2); pp. 185-202. Vitko, Jr., J.; Benner, R. E.; Shelby, J. R, (April 1982). Corrosion of Thin Silver Films in an Aqueous Environment. SAND82-8627, Albuquerque, NM: Sandia National Laboratories, “Corrosion Tests; Spray Tests with Various Sodium Chloride Solutions.” (April 1970.) DIN50021. Ber- lin, Germany: Deutsches Institut Fiir Normung e.V. (Available from Global Engineering Docu- ments, Santa Ana, CA.) Federal Specification; Mirror Inspection. (8 Septem- ber 1966).GGG-M-350a. FSC 5120. Interim Federal Specification; Mirrors, Glass. (5 April 1968.) DD-M-00411b (GSA-FSS).FSC 7105. Military Standard; Sampling Procedures and To bles for Inspection by Attributes, (29 April 1963).MIL- STD-105Da Santa Ana, CA: Global Engineering Documents; 64 pp. “Standard for Salt Spray (Fog) Testing; ASTM B117- 73.” 1983 Annual Book of ASTM Standards, Vol- ume 06.01. Philadelphia, PA: American Society of Testing and Materials, pp. 1-8. “Standard Practice for Performing Accelerated Out- door Weathering Using Concentrated Natural Sun- light: ASTM E838-81.” 1983 Annual Book of ASTM Standards, Volume 12.02.Philadelphia, PA: Amer- ican Society for Testing and Materials, pp. 595-607. “Testing of Materials, Structural Components and Equipment Method of Test in Damp Heat Alternat- ing Atmosphere Containing Sulphur Dioxide.” (December 1963.) DIN 50018. Berlin, Germany: Deutsches Institut Fiir Norrnung e.V,; 4 pp. (Avail- able from Global Engineering Documents, Santa Ana, CA.)

54 Sllver/Glass Mirrors for Solar Thermal Systems Chapter 5 Solar Mirror Manufacturing Methods

The future use of solar energy for residential, com- ture, and a silvered reflector surface will represent a mercial, and industrial applications depends to a mass-produced solar collector in the year 2000. large extent on the initial costs and the levelized costs of the total solar energy system. These costs, in Float Glass Manufacturing Process turn, will depend upon the economy and efficiency of mass production of solar mirrors and reflectors. The float glass process was introduced by Pilkington Solar mirrors and reflectors are manufactured mostly Brothers of England in 1959 and has revolutionized in either small quantities or in custom designs. The the flat-glass industry. Until that time, quality glass largest single quantity of solar mirrors manufac- was made by a plate process that included grinding tured by the glass and mirror industry so far has been and polishing the surfaces. This process is inherently the 93,000 m2 (1,000,000 ft2) of heliostats for the Solar slow, wasteful of raw materials, and requires a great One Solar Thermal Central Receiver Pilot Plant at investment in grinding and polishing equipment, Barstow, California. The manufacture of flat glass by the float process The glass for the heliostats was manufactured by the produces a continuous ribbon of high quality glass. Glass Division of the Ford Motor Company using a The process consists of (1)weighing and mixing the float glass process. The mirror preparation process proper ratio of raw material ingredients, (2) melting was performed separately by Gardner Mirror Corp. the batch raw materials in a high-temperature fur- under contract for Martin Marietta Aerospace Corp. nace, (3) forming the molten glass into a sheet by This two-phase process is typical of the mirror floating it on molten metal, (4) annealing the sheet, industry today for glass mirrors; that is, typically, (5)cutting the glass to size, and (6) packing. Figure the mirror superstrate or substrate is manufactured 5-1 is a schematic of the entire process. by one plant and the preparation of the mirror surface is performed by another. Therefore, the mirror The batching process involves weighing and blend- superstrate or substrate must be packed and shipped ing the raw materials of sand, soda ash, limestone, to a second facility to complete the fabrication. It is dolomite, salt cake, crushed coal, and rouge into a likely, however, that greater development of the homogenous mixture and delivering it to the melt solar industry will eventually necessitate the inte- furnace. gration of the two manufacturing processes in order to achieve economy and efficiency. The melt furnace is a large refractory furnace capable of temperatures in the range of l65OoC. The fur- The remainder of this chapter provides, first, a brief nace actually performs three functions. The initial description of the float-glass process used by Ford phase melts the raw materials into a viscous liquid. Motor Company; second, a review of the mirror Then, the melt proceeds through a refining zone industry manufacturing process today; and, third, a allowing gaseous inclusions (bubbles), generated projection of the solar mirror industry of the future. during the melting process, to escape from the molten glass. Finally, the molten glass enters a temperature- The description of the float glass process is excerpt- conditioning zone that stabilizes the glass at a ed from a paper by Goodyear and Lindberg (1980). desired viscosity. The presentation on the silver mil‘ror fabrication process is from, Heliostat Mirror Survey and Analy- Glass is formed into a sheet by flowing molten glass sis, prepared by Lind et al. (19791. The discussion of onto a bath of molten tin. A density differential the solar mirror manufacturing plant of the future between the glass and tin results in the glass “float- assumes that the superstrate/substrate fabrication ing” on the tin and forming a continuous ribbon of will be performed in the same facility as the process glass with essentially flat, parallel and fire-polished for preparing the mirror reflecting surface. The dis- surfaces. The glass ribbon is pulled along the bath of cussion is excerpted from, Preliminary Definition molten tin while simultaneously being cooled to a and Characterization of a Solar Industrial Process temperature at which it solidifies. Tin has a lower Heat Technology and Manufacturing Plant for the melting temperature than glass and remains molten Year 2000, by Prythero and Meyer (1979). The throughout the process. A reducing atmosphere is authors assume that parabolic trough collectors necessary in the forming area to control the oxidation made with a thin glass substrate, a support struc- of the molten metal.

Solar Mirror Manufacturing Methods 55 - Soda Ash J 1,

y_ Salt Cake -1h ”:....-

Refining Melti ng Conditroning Forming Cooling Annealing I I I I I I I I 4 J 000 I

Furnace

C Figure 5-1 Schematic Drawing of the Float Glass Manufacturing Process (Lind et al. 1979)

It is possible to stretch the glass ribbon during the hydrazine and its related salts. Further reducing- forming process to achieve the desired thickness solution improvements became available in the early before cooling to the point of solidification. The 2960s when formaldehyde-dextrose solutions were stretching process has been developed to a point that patented. These solutions and a method for continu- any thickness specification in the range of 1.5 mm to ous processing became almost universal for domestic 12 mrn can be met routinely. mirror production by the early 1970s. The copper-deposition process has experienced a The glass then proceeds continuously through the more pronounced change. Galvanic deposition tech- annealing lehr that alleviates stresses within the niques were used until the 1950s with early applica- glass ribbon and further cools it to below 15OOC. tions requiring mirror-dipping and later use allow- Then the glass ribbon is moved under a series of ing a conveyor belt motion. Patents filed in 1953 and cutters that cut it, “on-the-fly” to the desired size. 1954 introduced solutions for the spray application After the cut sizes are conveyed through a washing of copper employing copper sulfate solutions and operation, they are ready for packaging. metallic dust as a reducing agent. Variations on this generic approach are common today using iron filings. These solutions are suitable for use on a Current Mirror Production Processes continuous-process production line, Overview Today the major chemical outlets market competi- The evolution of mirror production techniques has tive forms of both the silver and copper solutions. been reviewed by Schweig (1973) who concluded The products are used interchangeably by the mirror that the generic chemical silvering solutions and manufacturing firms. Purchases are based on cost reactions used today have been available since 1876. and a desire to avoid a dependence on only one sup- Significant advances have been made during that plier. Differences in chemical performance or process time in the solution constituents [variety and amount) reliability are not mentioned by the coaters. to improve the film quality, deposition rates, and suitability for mass production. Early solutions re- Modern mirroring production lines provide for con- quired 2-5 minutes to deposit a silver layer and were tinuous production and are based on a horizontal- suitable for batch production using stationary or moving conveyor system using turning rollers. Indi- rocking tables with the solution covering the glass. vidual glass sheets are fed into the front of the system and no further handling occurs until the fin- Spray application was introduced in the 1920s, but ished mirrors are removed from the line output. A slow deposition time still limitsd production. Speedy typical production sequence is shown schematically silver deposition became available in 1940 when in Figure 5-2. The primary steps are indicated in Peacock introduced reducing solutions based on sequence from left to right.

56 Silver/Glass Mirrors for Solar Thermal Systems Id

Figure 5-2 Wet-Chemistry Mirror Production Sequence (Lind et at. 1979)

Although all manufacturers visited during Lind’s that will survive the chemical environment and the (1979) survey operated this generic sequence, there production temperature sequence without degrading were notable differences in their production lines. the mirror surface. Line widths ranged from 84 in. Many of the differences were related to the age of (2.1 m) to 130 in. (3.3 mf. Most conveyor systems equipment and the willingness demonstrated by the were in excess of 100 ft in length with the newer lines respective firms to upgrade their equipment. even longer. The trend to longer lines allows improved line speed and product throughput. Common line The basic configuration and physical layout of the speeds were in the range of 9 k 3 ft/min. (2.8 k 0.9 production line are determined at the time of the m/min.). The size of sheets that can be handled with original equipment purchase. The design of the line is ease ranged up to 96 in. x 208 in. (2.4 m x 5.3 m). generally a variation of a design recommended by the equipment manufacturer and modified by the pur- chaser’s technical and operating staff. A new line Glass Loading and Scrubbing thus represents the current industry status as determined by the equipment supplier, yet the line will The production sequence begins with the transfer of incorporate any advances or component preferences individual glass sheets from shipping cases to the of the coaters, based on their own experience. This conveyor line. Small-to-moderate size stock is trans- mode of operation allows the equipment manufac- ferred to the line by hand with all personnel wearing turer to benefit from many of the advances being gloves. Some firms have vacuum cup/mechanical made by their customers without having to perform handlers available for use with large sheets of glass. all the development effort. This gain is then passed Care is taken to load float glass with its untinned along to the current generation of customers. (air)surface up to receive the mirror coatings. This is done because production experience has shown a higher probability of good mirror quality when the Conveyor System untinned surface is coated. Incorrect loading will increase the chances for poor visual performance and The conveyor system is the most stable part of a lower silver/glass adhesion. production line. Most firms visited by Lind had upgraded selected areas of their equipment (scrub- It was also suggested that mirror quality could be bers, dryers, paint applicator, etc.) but had main- improved by specifying that glass be paper-packed tained the conveyer system essentially as purchased. rather than powder-packed. These terms describe This allowed the modular improvement of equip- alternative techniques available to glass rnanufac- ment with minimum downtime and no building turers to ensure the separation of adjacent glass changes required to house a larger system. Most con- sheets during shipment and storage. Failure to pro- veyor systems observed were fabricated from power vide this separation can frequently result in adjacent and idler rollers. Other systems use flat belts, waffle- sheets bonding to each other. Attempts to separate type belts, or cables covered with a resilient material them can damage the surface of the glass.

Solar Mirror Manufacturing Methods 57 Powder-pack is the industry standard and is accom- A related scrubbing mechanism is termed “blocking.” plished by spraying a light powder coating on the This employs felt pads on heavy blocks that are glass sheet just before it is taken from the production moved back and forth across the glass surface. An line €or packaging. Some powders contain adipic acid abrasive slurry is used to enhance surface cleaning. to reduce glass staining during shipment. Prelimi- Blocking has gone from the most common cleaning nary evidence indicates that standard glass cleaning system to relatively little use because of damage to the procedures are not totally effective in removing the glass surface from contaminants that become embed- adipic acid from the surface. Related studies are now ded in the felt pad. Blocking is being re-examined under way to determine if residual adipic acid is a with new slurries to see if it can reduce problems possible source of degradation in finished mirrors. as socia t ed with incornple t e surf ace scrubbing. An alternative packing method uses a sheet of paper The slurry residue is then rinsed from the surface between the surfaces of each adjacent sheet of glass. using water nozzles directed against the direction of This prevents bonding and does not require the use of glass motion (Figure 5-4). This moves the residue adipic acid. This does require additional labor and material for packaging and thus will generally cause an increase in cost. Also, the use of low grade packing paper may introduce similar problems associated with the acidity of the paper which remains from the milling process. The mirror-coating procedure is a wet process from the initial glass cleaning until the deposition of the metal layers is completed. The top surface is not allowed to dry in order to minimize the detrimental effects that would result from the deposition of solution residues on the surface. Figure 5-4 Water-Rinse Mechanism Using Directed Water Spray to A typical glass-scrubbing unit is shown in Figure Rinse Process Residuals Off Rear Edge 01 Glass (Lind 5-3. The scrubber employs an abrasive slurry applied et al. 1979) to the glass by small flow nozzles that move across the conveyor line transverse to the direction of glass toward the trailing edge of the glass where it finally motion. The transverse speed is coupled to the drops into a recovery tray under the conveyor rollers. conveyor line speed to ensure that the surface is Incomplete removal of the cerium oxide slurry wilI adequately cleaned. Cerium oxide slurry is the most generally result in a milky appearance of the depos- commonly used abrasive. Wetting agents are gener- ited silver layer. The rinse water is deionized to pre- ally added to the slurry to improve cleaning. Surface vent contamination of subsequent coating solutions abrasion is accomplished by rotating brushes or and reduce the chances of staining or clouding the pads that actually scrub the surface, The rotating metal layers. The use of deionized water is continued brush assemblies oscillate back and forth across the from this step through the deposition of the copper glass to provide uniform cleaning. Municipal-grade layer. Significant differences were noted in the min- water is used as the carrier for the abrasive slurry. imum resistivity values accepted by the different companies for their deionized water. The lowest value cited was 0.5 MWxn, while others were an order of magnitude larger.

Slurry Nylon Brushes Sensitization and S ilveri ng or Foam Pads The glass surface is then sprayed with a sensitizing solution that hastens the silver deposition rate and improves the silver/glass adhesion. The use of stannous chloride was first introduced in 1876, and it is still the most common sensitizer used for domestic mirror production, although palladium chloride is used occasionally. The sensitizer solution is sprayed onto the glass by a moving nozzle that oscillates back and forth across the conveyor line (similar to that used for CeO, as seen in Figure 5-3). Although the role of the sensitizer is not understood completely, it is thought that the resulting tin sites serve as nuclea- Figure 5-3 Abrasive Glass-Scrubber Using Continuous Slurry Ap- tion centers for the silver-layer deposition process. plication and Oscillating Assemblies of Rotating Brushes (Lind et al. 1979)

58 Silver/Glass Mirrors for Solar Thermal Systems The sensitizer solution is then rinsed thoroughly solution to the glass. It is estimated that 85%-90°/0of from the surface using deionized water through a the silver sprayed onto the glass deposits into the spray bar (Figure 5-4). Poor rinsing will cause a poor reflective layer. The process also provides improved quality mirror. A light water spray is applied to the silver/glass adhesion. glass to maintain the moisture layer and facilitate the spread of the forthcoming silver solutions in a uni- The chemical suppliers market the three component form layer on the glass. solutions in chemical concentrations that require k1:1 mixture on the glass. This is accomplished reli- The silvering chemicals are then sprayed onto the ably by using metered pumping systems for each sensitized glass. The commonly used chemical sys- solution to reduce product variability and simplify tems consist of three distinct solutions: (1)silver, (2) the required production adjustments. The metered

caustic, and (3) reducer. The chemical reactions that pumping systems provide an airless spray that redu- 1 result in the precipitation of a silver layer begin ces problems of overspray and chemical cross-talk when the three solutions are mixed. This reaction is from step to step. Production quality metered pump- localized to the glass surface by using a spray appli- ing systems can be purchased from at least one of the cator that simultaneously sprays the solutions at a chemical supply houses. common target point on the surface (Figure 5-5a). Mixture and initiation of the reaction occurs only on The silvering solutions are rinsed thoroughly from the surface of the glass. the glass to terminate the silver reaction and to prevent the entry of residual silver solutions into the All manufacturers use multiple sets of spray nozzles copper deposition portion of the line. Such chemical spaced along the direction of glass travel to renew contamination is expected to degrade the quality of the chemical solution periodically and produce a the copper layer. thicker layer of silver, The transverse spray assembly speed, conveyor speed, and nozzle spacings are selected to ensure that each region of the mirror is sprayed generally 6 to 12 times to provide the Copper Deposition and Drying required film thickness and uniformity. This coverage is generally obtained with four or five sets of Copper deposition occurs next and commonly em- spray nozzles and a transverse nozzle speed that ploys an iron filing/water slurry and a copper solu- allows approximately 12 strokes/minute across the tion. The precipitation reaction begins as soon as the mirror line. two solutions are mixed. The solutions are sprayed from an oscillating transverse nozzle system (Figure The most commonly used silvering chemical system 5-5b). The surface is first sprayed with the iron slurry utilizes formaldehyde and dextrose solutions as two to allow it to spread across the silver layer before the of the separate component solutions. This system copper is introduced. The copper solution is then was commercially developed in the late 1960s and sprayed simultaneously with another iron-filing achieved general acceptance by the early 1970s. It slurry. All manufacturers use multiple sets of double has the advantages of being a fast chemical reaction nozzles to reach the desired copper thickness and (suitable for fast production rates) and being a film Uniformity. An air sprayer is typically used for highly efficient reaction for silver transfer from the chemical application.

1 Figure 5-5 Reciprocating Spray Chemical-Applicatorfor Wet-Chemistry Mirroring Solutions. Specific nozzle arrangements are shown for a galvanic copper and a three-part silver process. The details A and B (on right) represent the norztes placed at A and B on left of illustration. (Lind et al. 1979)

Solar Mirror Manufacturing Methods 59 An alternative chemical system is now available for the reflective nature of the metal layers. This would copper deposition that does not employ an iron-filing tend lo create significant thermal gradients within slurry. It has the potential advantage of reducing any the mirror metallization that could degrade the final copper layer defects that may be caused by the gran- adhesion behavior of the metal layers. Heating from ular nature of the iron filings in the conventional the glass side is also thought to drive the moisture process. This uses a metered pumping two-component away from the silver/glass interface rather than application and airless spraying. The copper solu- toward it. tions are then rinsed completely from the glass with deionized water using a spray bar. ‘The metal layer drying also reduces the chance of problems that could occur with the application of an The deposition of the metal layers is now complete oil-based paint on a wet surface. Dryers were ob- and the procedure ceases to be a water-based pro- served that employed either resistively heated radiant cess. An air knife (Figure 5-6) is employed to remove ceramic cups or quartz tube heaters. The differences in these devices will impact operating costs, dryer

~ ~ box sizes, and maintenance requirements. No evidence was given to indicate preference for either type High based on the curing mechanism as long as they raised the mirror temperature to the same levei. Some manufacturers have suggested that optimal metal curing will occur when the layers are near 10OoC (212OFj.This is significantly higher than the typical values of 40°-550C (110°-1500F)used in the industry today. The actual temperature used must be low enough to allow good paint adhesion in the next process step and should not initiate silver-to-glass Figure 5-6 Air-Jet Drying Station Using Directed Air Flow to Force debonding as shown by Masterson (1983) and Lind Process Residuals Off Back Edge of Moving Glass (1982). Sheet (Lind et at. 1979)

water from the coated surface of the glass before Back Painting attempts are made to drive residual water from the metal layers. Visual inspection of the surface, after One paint application method predominately used the air jet dryer, revealed no evidence of standing for the production of quality domestic mirrors is the water or stray droplets on the copper. “curtain coater.” It is seen in a cross-section view in Figure 5-7(A).A thin curtain of paint is created that The mirror is then heated with infrared radiation flows continuously from the adjustable knife edge from the uncoated glass side to cure the metal layers drain in the upper reservoir. As the mirror passes partially by driving residual water from them. Heat- through the curtain, a uniform paint layer is depos- ing from the glass side raises the temperature of the ited. Roll coaters [Figure 5-7(B)] are still used for entire mirror structure. Heating from the coated side some applications but are not preferred on the basis would not heat the substrate effectively because of of product quality. The roll coater transfers paint

- ,.. . Upper -- .:- I. 1...... -.. Reservoir .’ .* , . . . - .‘L . I ,, i ..: -.

Figure 5-7 Cross-Section View of Two Types of Paint Applicators (Lind et al. 1979)

60 SilvedGlass Mirrors for Solar Thermal Systems from a reservoir to the mirror with a series of roller- dard line speed would produce very thick paint films to-roller exchanges. This is similar to the ink transfer for easily controlled paint flow rates. The compro- mechanisms used on many printing presses, Roll- mise solution uses practical paint flow rates and coated mirrors tend to exhibit fairly non-uniform accelerated mirror transfer through the curtain coater. paint layers with corrugations seen in the paint that This is illustrated in Figure 5-9. are reminiscent of the texture left by a household paint roller. Typical roll-coated paint exhibits a 50% The variable speed rollers are accelerated when a thickness variation associated with the corrugated pre-selected light beam is broken by the mirror’s pattern. presence. The section continues at high speed long enough for the mirror to clear the paint curtain. The Curtain-coater operation is further illustrated in roller section then slows down to transfer out the Figure 5-8. Paint is pumped into the upper reservoir. painted mirror and to load the next mirror. The

(A) Curtain Coater Between Applications Paint Reservoir with Knife Edge Drain From Recirculation System Fbxed Wires at Curtain Ends to Prevent Breakup

Recirculation System (5)Curtain Coater During Application Paint Reservoir With

From Recirculation System

Faint Recovery Tray To Recirculation System

Figure 5-8 Curtain-Coater Operation: (A) Between Applications, and (B) During Application (Lind et al. 1979)

The curtain then results from the gravitational feed photo-detector location and high-speed operating of paint through the adjustable knife edge drain time are selected on the basis of the mirror size being which allows the adjustment of both curtain uniform- coated during a given production run, Typical speed ity across the line and the nominal curtain thickness, changes required are from 6 to 8 ft/min (2.8-2.5 m/min) This type of feed system requires reasonable control on most of the line to 18 ft/min (5.6 m/min) through on the paint viscosity and its particulate content, the curtain coater. Most firms filter the paint supplied to the upper reservoir. A fixed wire is mounted at each end of the knife edge drain to serve as a guide for the end of the The alternative paint application techniques not curtain. This prevents the breakup of the curtain only affect the Iayer quality but also the painting ends that would normally occur. The edge wires costs. Spray painting [an early technique] required extend into the lower paint recovery tray. The paint the use of approximately 10 gal/ft2 (4.0 m3/m2) and from the recovery tray is filtered and pumped back to additional paint was lost because of overspray prob- the upper reservoir. Between mirror sheets the paint lems. Roll coating generally is done at approximately simply circulates through the system. When a mirror 11 gal/ft2 (3.6 m3/m2) to ensure that even the thin passes through the coater, most of the paint is inter- regions provide good mirror protection. But the cur- cepted by the mirror, This requires an automatic fill tain coater can provide good mirror protection at system for the reservoir in the standard production approximately 8 gal/ft2because of its excellent uni- process. formity. The net effect of these differences in required thicknesses and paint wastage represent a signifi- Practical implementation of a curtain coater also cant difference in painting costs. Estimated costs per requires the use of a variable speed section in the square foot (1976) were roughly $0.022, $0.015, and glass conveyor system to achieve the paint layer $0.024 for roll coating, curtain coating, and spraying, thickness desired. This is required because the stan- respectively.

Solar Mirror Manufacturing Methods 61 which removes any metal deposited by overspray. A (A) All Rollers are Slow During Loading variety of solutions are used but they fall into the Curtain Coater generic category of aluminum desmutters. Light Reservoir 4 Source Top and bottom surfaces of the mirror are then Paint Curtain cleaned before removal from the conveyor. Rotathg &-’& Recovery Reservoir cylindrical brushes may be used to scrub the glass (B) Mirror Breaks Light Beam, High Speed surface with the weight of the mirror holding the Rollers Accelerate glass on the brush. Another approach employs a water spray. Both surfaces are then dried with air 0 knife blowers. Most manufacturers then roll-stamp the back of the mirror with their name and the pro- (C) High Speed Rollers Move Mirror Quickly duction date to allow product identification. T h rough Pa int Cu rt ai n * On-line product quality assurance is limited to visual inspection for pinholes and blemishes. Mirrors -00. “a-... are turned over mechanically before attempting front surface inspection. Some firms employ front- and (D)High Speed Rollers Decelerate After Preset back-lighting with an observation point over the Time moving mirror. Others simply use overhead lights and observation from the perimeter of the production conveyor.

Removal from the conveyor system is done by hand (E) All Rollers are Slow During Unloading for small and moderately size mirrors. Mechanical handlers are available at some firms for use with large mirrors. Mirrors are repacked in’the original glass-shipping crates. Small felt pads are applied to Variable Speed Rollers the front surface of the mirrors to prevent damage to 0 Fixed Speed Rollers the finished product during shipment.

Figure 5-9 Production Sequence for Variable-Speed Roller Sec- tion Associated with Curtain-Coater Operation (Lind et al. 1979)

Typical Production Line Temperature Profile The mirror then passes through a vent hood that 3001 employs a circulating air sweep across the mirror to Ambient 1 remove solvents from the paint before entering the 250 paint dryer. The mirror temperature (residual from p+ 200 the metal dryer) and the forced air flow remove the 2 150 bulk of the solvent. The use of preheated and ambient Q) airflow for this “flash-off” process was observed, The fumes were exhausted from the production area, t- 50

I I 1 I 1 I I1 Final Drying, Cleaning and Packing Prefer red Produ c t i on Li ne Tempe rat u re Prof iI e The painted mirror then passes through an infrared dryer to dry the paint to allow handling, packing, and storage. Typical baking times are several minutes, and temperatures are near 125OC (250OF). This is not intended to cure the paint fully. The ideal drying temperature depends on the paint formulation, and the paint manufacturer’s recommendations should be heeded to provide optimum product quality. How- ever, a practical temperature may be governed by the production line’s capability to cool the finished mirror for final packing without thermally inducing breakage. Figure 5-10 shows the comparison between the typical production temperature profile and one designed for the supplier’s recommendations.

The mirror then rolls across several rollers that carry Figure 5-10 Production-Line Temperature Profiles (Lind et al. a chemical solution to the uncoated glass surface, 1979)

62 Silver/Glass Mirrors for Solar Thermal Systems The operating speed for the integrated praduction manufacture parabolic trough support structures it line must be selected to maximize the output volume would be advantageous to manufacture all specialty while maintaining the product quality. The speed items within the factory to allow for a well-planned may be limited by mass production process. Deposition times required for silver or copper. Based upon the actual design of the collector, it Desired glass temperature profile along the line. would require several different special components and materials, These include a glass reflector, sheet Power rating of infrared dryers. molding compound, black-chromed absorber tubes, Component spacing. sun-sensors and controls, and the tracking drive Heat capacity of the glass substrate [proportional mechanism. All these items would require an ex- to thickness). panded manufacturing operation unlike present solar manufac turing operat ions. Glass handling rates for loading and unloading. From an analysis of the manufacturing techniques Several examples of these limitations should clarify required, it appears that the glass making, silvering, the possible interactions. The use of thicker glass and mounting to a parabolic trough shape should be substrates will require the use of slower line speeds. carried out in one continuous process to avoid exces- The increased heat capacity of the glass will require sive handling. There are two options available. One longer times to heat and cool during the processing. is to install a glass-making operation in the collector Therefore, the speed must be reduced to maintain the plant itself. The other is to make the parabolic same temperature profile along the line. Component trough, send it to a glass-and-silvering operation spacing and required process times (e.g.,silver spray- elsewhere, and then ship it back for final assembIy. rinse separation and silver deposition time) may With present shipping procedures, it does not make impose a maximum line speed compatible with main- sense to ship the glass to the collector plant once it is taining product quality. Typical production line made, because of the potential for damage. There- speeds vary from 7 ft/min to 12 ft/min for the differ- fore, the solar plant in this study is characterized ent firms. The potential of reaching 18 ft/min with an with a glass-making and silvering operation within optimized line configuration was mentioned. the plant. The manufacture of black-chromed absorber tubes could also be conducted in the solar plant. The prin- Manufacturing Plant in the Year 2000: cipal reason for including the chroming operation is An Overview the volume of electroplating required. Other components that could be manufactured in the plant are the A plant manufacturing parabolic trough collectors sun-sensing components and the control systems. would have to be a large operation to be profitable The operation would consist mainly of assembling and to produce a well-priced unit. A plant producing electronic components into specialized systems for a minimum size of 1 million ft2of collector may only use with the collectors. be able to assemble the collectors, with most of the materials and components being manufactured else- Hence, the solar manufacturing plant is defined as where. The larger the plant, the more likely it is to manufacturing the glass, silvering it, and attaching it conduct several manufacturing processes. A plant to a plant-manufactured trough. The absorber tubes capable of producing 10 million ft2of collector would would be black-chromed and installed in the fabri- certainly contain several manufacturing operations. cated collector unit. The electronic controls and sun-sensing devices could be assembled at the plant. It is likely that a plant producing parabolic trough The collector units would be assembled to the extent collectors will also produce other types of solar col- that shipping is practical; subsystems and compo- lectors rather than the other components used in a nents would be packaged with the collectors for solar system. For example, a plant can manufacture distribution. reflective mirrors for different concentrating collectors and heliostats. Since the same types of technolo- The only packaging of a “system” to be carried out at gies are involved, it would make economic sense to the plant would consist of the modular collector manufacture similar items. A plant operation is more arrays with all appropriate components to make likely to manufacture similar products rather than them functional. Additional specialty components products used in total solar systems that are dissimi- needed for the complete solar system could also be lar in construction. packaged. The plant characterized in this study is considered to be a large capacity plant that manufactures parabolic trough collectors. It could easily add other Manufacturing/Assembly Processes for types of line-focusing and point-focusing collectors Major Components to its production activity. A large operation would benefit from having most of the collector manufac- Reflective material. The reflective material is a thin turing conducted within the plant. For example, to glass about 1.5 mm thick, which would be rnanufac-

Solar Mirror Manufacturing Methods 63 tured at the plant, silvered, and cold-pressed onto the inspection of a completed component: parabolic trough structure in a continuous in-line 1. Dry ingredients are mixed in a hopper equipped process. This glass-manufacturing process is similar with an agitator. to other types of glass-making operations. It is energy intensive and requires a considerable amount 2. The mixed compound is sent through a continu- of area for the kilns. ously mixing roto-feed extruder where the resins and catalysts are mixed with the dry material. The process involves seven steps: 3. This material is sent to the sheeting equipment 1. Receiving raw materials for glass. where glass fiber is chopped and forced into the 2. Batch-mixing of the materials for firing. resin. It is impregnated, compressed, and kneaded 3. Remelting, melting, and heat refining in a gas- into a sheet about 0.1 in. thick, which is rolled up fired tunnel kiln. for use in the molding operation. 4. Forming by drawing the glass downward in a ver- 4. The sheets are cut to size for the shape of the tical fashion to obtain sheet glass (or by a fusion parabolic trough mold. float-glass process). 5. The sheets are placed in the mold in sufficient 5. Sending the sheets of glass through an annealing quantity to achieve the proper thickness of plastic kiln (lehr)to reduce the glass temperature to room for the component. temperature. 6. The molding operation uses a match die molding 6. Cutting the glass to appropriate lengths. machine in which heat and pressure are applied to the mold: heat at about 17OoC with pressure of 7. Finishing as necessary by trimming and polishing. 500-1000 psia is required, In subsequent operations, the glass is silvered and a 7. The trough is removed, trimmed, and inspected. protective coating is placed on the back. The silvering process involves several steps: Absorber assembly. The absorber assembly consists of a steel tube that has been black-chromed, a 1. Cleaning of the glass. plugged inner tube, a Pyrex glass outer tube with 2. Rinsing. reflector coatings and appropriate gaskets, fastening devices, and support brackets. Most of the materials 3. Sensitizing the surface with a tin solvent. would be manufactured elsewhere and assembled at 4. Rinsing. the plant. The black-chrome plating process is an 5,6,7. Spraying (three stages] the silvering com- optional manufacturing operation which would be pounds on the glass. carried out in the plant if the volume is sufficient. 8. Rinsing. The manufacturing operation associated with the 9. Copper plating the silver using a galvanic process. absorber support bracket, bearing housing, and trough attachment assembly would require several 10. Rinsing. operations. The operations include metal cutting, 11. Drying. welding, drilling, machining, and fastening. Mate- 12. Spraying a protective coating on the glass (possi- rials used would include steel tubing, angle and plate bly a plastic-based material]. stock, bearing housings, and fastening devices. Finally, the glass mirror is cold-pressed into the The absorber tube would be pre-threaded and pro- premade parabolic trough and bonded to the surface. vided with appropriate fittings. Assembly of the gaskets and glass tube could be carried out at the The glass-making operation would require a process factory or possibly performed at the job site. line of about 100 ft to 150 ft in order for the process to be carried out in one operation. The silvering opera- The optional black-chroming process is an electro- tion would require another 75 ft to 100 ft of process plating procedure involving 11 steps: line. I. Sandblast the steel pipe to remove oxides and foreign materials. Parabolic trough support structure. The manufacture 2. Clean with an electrolytic chelating cleaner. of the parabolic trough would use a sheet-molding 3. Acid dip. compound, This compound is a reinforced plastic 4. Electrolytic chelating cleaner. containing resin, fillers, catalysts, thickeners, carri- 5. Acid dip. ers for the thickeners, and reinforcement material. Once mixed, the sheet-molding compound is in the 6. Rinse form of a dough-like mixture in sheet form. This 7. Nickel plate (dull]. material is placed in a matched-die molding machine. 8. Black chrome plate. 9. Water rinse. The manufacturing process for making sheet-molding compound parabolic troughs involves a process rang- 10. Alcohol rinse. ing from mixing raw materials to trimming and 21. Air drying,

64 Silver/Glass Mirrors for Solar Thermal Systems This process uses a series of dipping tanks and elec- Activities of the Solar Manufacturing Plant trolytic tanks and may require a process line of 50 ft to 60 ft. The solar manufacturing plant would carry out several major activities including receiving, warehous- ing, metal working, electroplating, glass making, sil- Collector support stand. The collector support stand vering, plastic forming, assembly and fabrication, is fabricated from low-cost steel. This manufactur- painting and finishing, packaging, and shipping. ing operation involves a basic metal-working proce- Subsidiary activities indirectly related to the manu- dure, Steel tubing and plate are cut, drilled, machined, facturing process would include management, water and welded to form the support stand. The stand treatment, air quality control, and solid waste consists of a mounting plate welded to the stand and disposal, the bearing house attached to the top end of the stand. The stand is cleaned, primed, and painted: Many operations in the manufacturing plant require then it is shipped to the job site where it is placed on that processes be placed in line with each other to concrete pads and the parabolic trough assembly is reduce excessive handling and additional space re- added. quirements. It will be necessary to have the glass- making and silvering operations together and they will require the most space in the manufacturing Tracking drive mechanism. The hydraulic-action plant. Many other operations would have to accom- tracking drive mechanism consists of components modate these operations. For example, the attach- such as actuators, piping, pumps, and control equipment of the glass mirror to the parabolic trough must ment, which would be assembled in the factory.It is occur at the end of the silvering process. Other opera- not anticipated that machining would take place to tions such as plastic forming and the assembly of the build any parts of the hydraulic drive in the plant. electronic controls can be carried out in remote areas However, the double-acting actuators might be con- of the factory. They need not be integrated into the structed in this plant. production flow of the collector assembly. The hydraulic components would be placed on mounting brackets attached to the parabolic trough General Description of the Manufacturing assembly and the support stand. As many of the components as possible would be attached to the Plant collector units in order to minimize on-site assembly operat ions. The manufacturing operations described in this study would require that the plant be located in an area zoned for heavy industry in a community or county, Sun-sensing device and control system. The sun- mainly because of the glass-making and electroplat- sensing device and control system could be manufac- ing operations. If these processes were not included, tured in the plant. If built by another manufacturer, the plant could be considered a light industry. the components could be shipped to the factory or The plant’s internal configuration and physical ap- delivered to the job site for attachment. Using the pearance would not be unusual. However, air pollu- solar heating and cooling industry as a basis, it tion control equipment and a water treatment plant would seem that the electronic control firms rather would be evident as additions to the building. Min- than the collector manufacturer would be the most imal outside area would be required for the plant. likely supplier of these components. However, if Protected storage areas for the glass raw materials these components are produced by this factory, it may be required. would be an assembly operation. The operation would include obtaining microprocessors, electronic Externally, the plant should change considerably parts, switches, and components from electronic compared to current configurations. Assuming the firms. The control systems and heat-flux sensing existence of a large solar system plant in the year devices would be assembled according to the require- 2000 implies that the use of solar energy will be ments of the collector tracking specifications. This significant. It is probable that the plant itself will use operation would be a labor-intensive, bench-top solar power extensively. The types of solar systems assembly operation. or elements could include hybrid passive and active space heating or cooling, high-temperature process heat for either preheating or direct process heat for Piping couplings. The collector array modules will the kiln operations, and wind power or photovoltaics require adequate coupling to the heat-transfer fIuid for electricity. Cogeneration concepts might be inte- manifold network of the collector array, which neces- grated into plant operations for additional electric sitates the use of either flexible or swivel-jointed power generation. couplings to allow for the movement of the collectors. Since these special couplings are manufactured by In urban settings it is unlikely that biomass could ie other industries, the solar manufacturer would pur- used as a back-up energy source. Passive energy con- chase these units and attach them to the collector servation techniques; for example, trees or struc- system. tures in strategic wind-breaking or sun-shading

Solar Mirror Manufacturing Methods 65 locations, the use of earth berms or subterranean construction, and absorbing or reflecting external wall surfaces, should be evident. Also, the plant configuration and orientation could maximize the avail- ability of insolation. It would be preferable to have access to a railroad and to a major highway. Because of the large quantities of materials required for glass making, it would be beneficial to locate the factory near these sources of raw material. References

Goodyear, J, I<.; Lindberg, Victor L. (1980). “Low Absorption Float Glass for Back Surface Solar Reflectors.” Solar Energy Materials, Vol. 3 [Nos. 1, 2); pp. 57-67. Lind, M. A,; Buckwalter, C. Q.; Daniel, J. L.; Hartman, J. S.; Thomas, M, T.; Pederson, L. R, (September 1979). Heliostat Mirror Survey and Analysis. PNL-3194. Richland, WA: Pacific Northwest Laboratory. Lind, M. A,; Chaudiere, D. A.; Dake, L. S.; Stewart, T. L. [April 1982).Electrolysis Coatings for Silvered Glass Mirrors. PNL-4257. Richland, WA: Pacific Northwest Laboratories: 36 pp. Masterson, K. D.; Lind, M. (April 1983) Matrix Approach for Testing Mirrors - Part 1. SERIITR- 255-1504. Golden, CO: Solar Energy Research Institute: 35 pp. Prythero, T. and Meyer, R. T. (November 1979). Pre- liminary Definition and Characterization of Q Solar Industrial Process Heat Technology and Manufac- turing Plant for the Year 2000. LA-8137-TASE. Los Alamos, NM: Los Alamos Scientific Labora- tory; 34 pp. Schweig, B. (1973).Mirrors; A Guide to the Manufac- ture of Mirrors and Reflecting Surfaces. London, England: Pelham Books.

66 Sitver/Glass Mirrors for Solar Thermal Systems